Boost the people. Own the judgment. Ship the proof.
Odoo work is compressing into prose and agents, and Odoo itself is climbing the stack: whoever owns the last mile owns the client. The firms that move first get to choose their own shape. This site is the whole idea in one place: why the shift is real and bigger than AI, what we already ship, the bets we make, the clock we run against, and who does what now. Geen slideware. Argue with it.
The ERP is not dying. The way we work on it is. That's the whole manifesto in two sentences. The long version is eight forces.
The arc: what changes (F-01 to F-03) → the squeeze (F-04) → the answer, two lanes (F-05, F-06) → shelter and opportunity (F-07) → the doorway and the prize (F-08)
Odoo stays. It becomes the clean layer underneath: the record, the data model, the permissions, the audit trail. That layer gets more valuable in the AI age, not less. Agents need a reliable substrate even more than people do.
Everything on top of it is moving. The clicking, the configuring, the module-by-module implementation grind: that work is compressing into prose and agents. The click-path knowledge that carried this trade for fifteen years is depreciating fast. The judgment about what to configure is not. That was always the part worth paying for. Now it's the only part.
ExhibitOdoo 19 ships prompt-based server actions · NetSuite AP automation went 7% → 29% in a year
F-02
The intelligence moves out of the ERP, into the agent
The sharpest line in this whole field came from a finance practitioner on LinkedIn, someone with no software background who wired Odoo to Claude Code and had an agentic finance flow running in about a day: the intelligence isn't in the ERP, it's in the agent you connect to it.
His build was a concept, not a product. Scaffold grade, an MVP at best, nowhere near production at scale. Doesn't matter: a non-engineer got there in a day. And we know the direction is right because we lived it ourselves, a real client payroll rebuilt in hours, reconciled to the decimal. Receipts one section down, in Proof. The build is cheap now. The judgment behind it is not.
ExhibitOur own payroll rebuild: ~3 hours vs 20-40 consulting hours. Full story in Proof
F-03
Code is a commodity. Judgment and taste are not
This is not koffiedik kijken. Odoo said it on stage, at Partner Day 2026, to all ~5,000 partners at once: code is becoming a commodity, and the market doesn't need integrators anymore. It needs business optimisers.
The vendor just told its entire channel to move up the value chain. Every partner heard the same sentence on the same day. Most will put "AI" on the website and change nothing. A few will actually move. The clock now runs for everyone.
And the same tool cuts both ways: used with judgment it makes you 10x, used without it, it produces confident slop at speed. That's why our posture is boost-don't-replace. The machine takes the execution, everyone keeps the judgment. A small, dense team with taste beats a big bench every time.
ExhibitPartner Day 2026 keynote, to ~5,000 partners at once
AI replaces execution, not judgment and taste. Hammers favour people with arms.
F-04
Odoo itself is coming up the stack
The vendor is not sitting still. Native AI ships in the product. Odev AI automates migration work. And with Partena, Odoo stepped straight into managed payroll services, its biggest contract ever. Odoo is moving up the stack, absorbing the commodity work its partners used to bill.
That thins the classic partner role from two sides at once: AI eats it from below, the vendor from above.
Set AI aside for a moment, because this one is bigger than AI. Every release, Odoo absorbs more of what partners used to bill: features that replace customisations, hosting, and now services and migrations too. That is the vendor's job, and it will not stop. So the core strategic question for any Odoo firm this decade is not an AI question at all: when the vendor keeps climbing, where do you live? The only durable answer is the client's operation. Their processes, their data, their people, their outcomes. The product can never own that, and the vendor will never send someone to run it. Whoever owns the last mile owns the client. AI decides how fast and how profitably you can get there. The next two forces are exactly those moves.
ExhibitOdoo 19 native AI · Odev AI upgrade tooling · Partena, Odoo's biggest contract ever
F-05
Everyone runs FDE at enterprise rates. Nobody at ours
This shift creates a role. The forward-deployed engineer, FDE: someone who owns a workflow's outcome in a client's live operations, not just its go-live date. Everyone big has noticed. OpenAI stood up DeployCo. Anthropic partnered with Blackstone. Every major SI launched a branded FDE program in the past year.
All of it enterprise-platform, all of it at enterprise rates. By the labs' own math a €30-80k engagement can never carry a $400k engineer, so they sell SMBs tokens and keep the humans for the enterprise. Not one of those programs operates at Odoo-SMB economics, the band where our clients actually live. That lane is empty. The first FDE-for-SMB studios are only now forming in the US, and none of them know Odoo or EU statutory work. The first firm to run real FDE work at those economics owns it.
ExhibitOpenAI DeployCo, >$4B committed capital · AWS stood up a $1B FDE org · FDE job postings +729% in a year. All enterprise-rate, none in our band
F-06
The second lane is width: more than Odoo, around Odoo
F-05 is the depth lane. This one is width. The proof band shows a full Android app and a complete website with client portal, built end to end with AI. Work that used to need a separate agency, a separate discipline, a separate quote from someone else. Now it's a project.
That changes what an Odoo partner can be. Everything a client needs around their Odoo (the app in the field, the portal, the headless website, the integrations) becomes work we can take on and bill, without hiring a second company's worth of specialists. We do it only for clients on Odoo and connected to their Odoo, so the focus holds. But the service surface widens, and every widening is revenue that used to walk out the door.
ExhibitThe Android app and the site + portal in Proof · built end to end with AI, no second agency involved
F-07
Europe's complexity is shelter and opportunity
Hundreds of millions went into AI-native ERP and accounting challengers in a single year. Nearly all of it builds for the US mid-market, and nearly all of it deliberately avoids what defines our clients: multi-country VAT, statutory filings, localisation per country.
That complexity is exactly why the funded wave won't simply roll over European SMB, and why depth here compounds instead of commoditising. The money is building around Europe's complexity. We build into it. And it cuts both ways: every statutory quirk the wave avoids is billable depth for the firm that masters it with AI. The complexity is not just cover. It is inventory.
And we lean into it deliberately: 360 is going EU-first on its own stack, and is already moving piece by piece to European providers. Two dependencies remain, Claude and Google, and those are the hard ones: there is no European frontier model yet. The moment a credible one exists we can switch, because everything around it already lives here. And the ground moves our way: EU inference capacity is building out fast, Cerebras alone plans 200MW across France and the Nordics by end of 2027. For clients that is not ideology. It is risk management they can point to.
ExhibitBasis at a $1.15B valuation · Rillet, DualEntry, Campfire $200M+ combined · all US-first, all localisation-light · our own stack: EU piece by piece, Claude and Google the two that remain
F-08
The month-end close goes first
If you want to see what ERP work looks like in 2028, watch what's happening to the month-end close in 2026. Every incumbent and every disruptor is converging on the same spot: close, AP, AR. NetSuite ships Autonomous Close. Rillet sells continuous close. The close is high-frequency, painful, measurable. The perfect first target for agents. The rest of ERP work follows the same curve, one to three years behind. The close is the canary.
And the canary guards a prize. If agents can run the close, you can set up and run an entire administratiekantoor on Odoo, AI-first: agents do the processing, people do the judgment, and the firm charges for outcomes instead of hours. Someone is going to build that firm. The strategy is about why it should be us.
ExhibitNetSuite Autonomous Close · SAP finance agents · Rillet continuous close. All converging on the close
The three focus areas, inside out
Why cut the plan this way? Because the only credible AI partner is one that runs on AI itself. We can't sell what we don't live. So the work starts with the people, then the firm, and only then the client. Practice what we preach, in exactly that order.
The order also de-risks the whole plan. The inner two rings need no buyer: boosting a colleague pays off the same week, and automating our own operation pays off whether or not a client ever signs. The outer ring is the only one that earns revenue, and the only one that depends on someone else saying yes. Building inside-out means the plan creates value even while the market makes up its mind. It's also why the market's favourite move (put AI on the website, change nothing underneath) is the one thing this plan cannot do.
So everything in this plan lives in one of three areas, and each ring earns the next. Every bucket, workstream and action further down carries its area tag, so you can always see which of the three a piece of work serves. Some work touches two. Each piece has one home.
1 · Individual team members
Every team member at 360 gets to use AI to its max potential, and closing the gaps between people is the plan's job, not the individual's. Everyone gets the tools, the skills and the time. Nobody gets left behind for lack of support, and nobody is forced into one shape. If we want to be the AI partner for our clients, we practice what we preach at the individual level first. No buyer needed, so it goes first. On the board: … actions
2 · 360 as an org
The org runs front-runner: anything in our own operation that AI can streamline, we build the capability to automate. Whether we apply it in the arbitrage phase (the stretch where we still price at classic rates while delivering at AI speed) is a pricing question. The capability is not optional. AI leader in our own org before we claim it for anyone else's, and the efficiency dividend banked as margin and quality. On the board: … actions
3 · The client side
THE AI partner for our clients. Strong focus on Odoo, sometimes beyond it. It starts with simple 360AI (our AI add-on inside the client's own Odoo) and evolves into FDE advisory, agents, autonomous workflows, sparring. AI-driven accounting lives here too, and so does the headless websites and portals track, built beside Odoo rather than on it. The only ring that turns into revenue, and it is fully driven by the two inside it. On the board: … actions
And that is why 360 moves now
The window is open and it has a date on it. Odoo just told 5,000 partners to move, and is climbing the stack itself. The head start is measured in quarters, not years. The enterprise FDE wave hasn't come down to our band yet, the width lane has barely been noticed, and the challenger money is building around Europe's complexity, not through it. The close is compressing this year, not someday.
The only time a firm gets to choose its own shape is while the old model still pays the bills. Wait for the customer signal and all that's left is triage.
So we move now. Best, not biggest. Deeper into the client, wider around their Odoo. And nuchter about it: no AI-transformation theatre, just work taken off people's plates, week after week, until the firm underneath has quietly become the thing this page describes.
02
Geen slideware
Proof, not promises
Every AI plan promises. This one shows first. Twelve proof points, all real, all checkable. And a thirteenth in the corner that collects your verdicts on the other twelve.
0.0002%
A real payroll, rebuilt in ~3 hours
A real client's ~53-employee payroll turned into an installable Odoo module in about three hours, while doing other work. Reconciled against the client's own Excel to 0.0002% per line. That job historically books 20-40 consulting hours. And the precision went up: the reconcile test is the definition of done.
~400 × €15/mo
360AI is already sold, not just built
360AI, our AI add-on inside the client's own Odoo, runs in production today with around 400 subscribers at €15 a month. Real product, real recurring revenue, before the flagship work even starts. This plan doesn't start from zero. It starts from a paying base.
~1 hour
A migration, unattended
Claude got a copy of a client database with custom code, on our own infrastructure, and one instruction: go 17→19, set up Docker, test everything, stop only when it's clean. It ran about an hour and finished, test suite green. One run, one database, so a signal, not a stat. Our read: ~80% of migration work coverable this way. Odoo's own Odev AI claims ~60%. The vendor confirming the premise for us.
14 Jul · kickoff done
A client already in motion
A client agreed by email, kickoff done 14 July. First agent: one invoice inbox, every invoice routed to the right entity, reusing a flow we already built for another client. AP first: the exact spot the whole market is converging on. Target: live 1 August for 2 of 4 entities. And they asked for the full agentic picture, not just a demo.
0 Excel files
NetSuite to Odoo, no spreadsheet in sight
A full migration flow that imports a company's NetSuite data straight into Odoo for a live implementation. No export-massage-import loops, no mapping spreadsheets, zero Excel work. Built once, reusable for the next system switch.
site + portal
A full website and client portal, AI-built
The complete website and client portal of Douwe's previous company, designed, customised and built with Claude end to end, including the SEO work and an AdWords strategy. Marketing sites and portals are agent work now.
offline first
A full Android app, Odoo-integrated
A complete Android app for a client whose people work offline in the field: local-first, syncs with Odoo the moment there's signal, built end to end with Claude. Native apps used to be a separate discipline. Now they're a project.
live today
The agent plug, live on our own Odoo
MCP is the standard plug that lets an AI agent talk to a system like Odoo. Ours, the safe-MCP (action-aware, OAuth), runs live on our own environment: read-only scopes enabled today, the action-whitelist layer already built. Own environment first, clients after.
< €5 / run
A full review for under a fiver
The version-delta harness, our automated code review for version upgrades: a full 18→19 environment checked for under €5 a run. Every ingredient already owned, ships in days. The cheapest visible win we have.
1 prompt
A demo environment for one screenshot
Ask how a customisation would look and Claude spins up a local Docker Odoo, builds the customisation, and screenshots the UI to show the client. One prompt, all of it. The screenshots feed straight into the documentation factory next door.
docs on demand
The documentation factory
One skill turns modular notes and screenshots into finished BRDs, training guides and walkthrough pages, built and deployed as polished documents. The doc work that used to eat evenings now takes a prompt.
built & running
Gijs's ticket factory
Tickets go in, worked results come out for review. Gijs built it solo, one of the pioneer projects this plan connects into the shared way of working. Ask him to show it.
argue mode · bottom right
The feedback loop, feeding on itself
Proof point thirteen is the button in the corner of this page. Hit Argue with it, click any block (or roll the dice and let it pick one), leave a one-line verdict. Everything files into a reviewable log, and changes fold back into the site with a note. You can file feedback about the feedback tool. Very meta. Have at it. Desktop-only for now: on a phone the button politely sends you to a bigger screen.
Eleven of the twelve demonstrable on request today, the last one ships this month. Ask and we'll show the receipts · internal page
03
The choices
The strategy: bets against the grain
Believing the shift is real costs nothing. Strategy is where it gets expensive: the places we deliberately do the opposite of the market, the four buckets where it becomes real, what all of it does to 360's own shape, and the risks we say out loud instead of burying in an appendix.
Best, not biggest.360's AI wins on depth, not headcount. Revenue per FDE, agent reuse, outcomes we operate. Those are the numbers we steer on, and every scale number is a side effect, never a target. When the build is as cheap as Proof shows, what compounds is everything around it: the operating loop, the trust to run a client's work, the vertical depth. F-04 said whoever owns the last mile owns the client. Depth is how we own it. Width is how everything around it stops walking out the door.
Six bets against the grain
The market
Sells AI as headcount reduction and prices the layoff.
360Boost, don't replace
10x the team we have. The 8 hours that drop to 4 become margin and quality, never someone's job.
Why
So far, the firms that cut people haven't outperformed the ones that didn't. Gartner found the same in 2026. And depth needs the team intact.
The market
Chases client AI demand with demos and pilots, assuming the buyer is ready.
360Internal-first, because the switch is hard
Grab the certain value inside first: cheap, fast, low risk, no buyer needed. Client demand is a test we run, not an assumption.
Why
The people who do the selling tell us clients love the demo but the final switch is hard. Nobody is lining up yet.
The market
Builds gateways, wrappers and chat plumbing.
360Own the domain layer, not the plumbing
Fork or buy the commodity gateway. Build the domain tools, evals, cost-routing and experience on top.
Why
Plumbing commoditises first. The Odoo-and-vertical depth is the layer point solutions can't reach.
The market
Sells autonomous end-to-end finance and quietly absorbs the risk.
360Automate without moving the liability line
Detect-and-propose, always checkpoint-gated. Config, advice and tooling are ours. The client's data and books stay theirs. We do not sell autonomous finance in 2026.
Why
The line is 360 culture, held independently by several senior voices. It's also what makes automation safe to say yes to.
The market
Bills hourly, so every AI hour saved is revenue lost. Or discounts first.
360Ride the fixed-price arbitrage, then rotate
The client gets a fixed price and cost certainty, the delivery risk sits with us. We price at market, deliver at AI-compressed cost, keep the spread, reinvest it. Then rotate toward hybrid outcome pricing before Odoo 20 flattens the surface: native AI shrinking the delta we can price against.
Why
Fixed price converts efficiency into margin instead of punishing it. We ran fixed price before and stepped away over scope discipline. So any pilot solves scoping first. That's why bounded, repeatable migrations are the cleanest candidate.
The market
Builds products for clients from day one, guessing at what they need.
360Build for ourselves first, sell what survives
The brain, the factories, the harnesses: built for our own daily use. Whatever survives thirty demanding insiders becomes a client offering, already battle-tested.
Why
Our own tinkering is the cheapest R&D and QA there is. The proof band is full of internal tools one step from being products, and the 360 brain itself is first in line.
The arbitrage, said out loud
The market is not asking for super-AI efficiency yet. Clients price the work the way they always have, and nobody demands we deliver at agent speed. That is the arbitrage: we price at market, we deliver ever cheaper, and the spread is ours. Why leave money on the table while nobody is asking?
So we get to choose. Some efficiencies we run internally and never sell. Some capabilities we build and deliberately keep on the shelf. Not running something commercially is a pricing decision, not a capability gap. The capability itself is never optional: that is the 360-org focus area, front-runner on everything.
And we watch for the flip. The moment the market starts pricing against AI delivery, or ideally just before (the signals in this plan exist to predict it: the POC, the Odoo 20 tell, the efficiency baseline), we press the button and GO. Everything switches to AI-speed delivery at once, because it was ready all along. The window belongs to whoever is ready before it opens.
The first place we press it: the 84 migrations to Odoo 20 waiting in our own install base. Two ways to do those. The old way: by hand, one at a time, at rates every competitor is about to discount. Or our way: build the migration engine once (scripts, the agent, a full test suite, automation aiming north of the ~80% we already measured), bill a fixed fee per migration, and keep the margin the engine earns. The client pays market price either way. The engine decides who keeps the difference. And the clock is real: Odoo's own Odev AI is coming for this same work (F-04), which is why the engine gets built now, while the margin is still a partner's to keep. The full pipeline lives in the roadmap.
The four buckets: where it becomes real
Org: connect the chapter, don't reorganise it
360 orgindividual
The AI work is fragmented across a handful of strong people, each working solo. It needs a convener, not a visionary.
The glue is the 360 brain (three repos, broken down in the roadmap) plus a consented intake skill. Manual knowledge bases die on the contribution problem, so contribution gets automated: harvested with consent, preview before send, never client data.
We measure adoption to boost people, never to audit them. The baseline survey is framed as show me what works for you, and the results go back to the team within a week. And we finally instrument the efficiency gain, because nobody owns that number today: without it, whether work is genuinely faster or just pulls in more tickets stays a guess.
When a job drops from eight hours to four, the dividend is not a pure cost cut. The reclaimed time goes into listening better and implementing better, captured as quality and margin. And the org's own stack goes EU-first too, provider by provider: the F-07 move made real in our own operation.
Products: one flagship, deep
client
The pick is finance, accounting-with-AI: close, AR, AP. The bench is strong, and the destination is bigger than a feature set: set up and run an AI-first administratiekantoor on Odoo. Agents do the processing, people do the judgment, the firm charges for outcomes. This is the most funded lane in the market, and that is exactly why F-07 holds here hardest: the funded wave builds localisation-light, and our clients live in the complexity it avoids. So the near-term flagship is governed, localised finance automation for EU SMBs on Odoo, not generic AI bookkeeping, every slice a piece of that firm, always behind the audit gate.
Around it, two wedges. The safe-MCP: Odoo 20 ships a native MCP, read-only by default, with no human-in-the-loop once writes are enabled. Access itself becomes free, so the product is the governance on writes: approvals, scoped permissions, audit, rollback. And migrations, the product with a date on it, running as its own workstream in the roadmap.
Plus the width track from F-06: apps, portals and headless websites for clients on Odoo, taken on the moment a client asks, billed as normal projects, no second agency. And underneath it all, Bet 6: what we build for ourselves becomes what we sell. The doc factory, the review harness, the brain: every internal tool that survives our own daily use is one step from a client offering.
Clients: make demand, don't wait for it
client
Both real buyers we've seen only showed demand after we showed them something. The tool is the production-data POC: a working agent on the prospect's own data, before any contract. Always under a signed NDA and data-processing terms, sandboxed, never used for training.
And we find that demand transparently. We tell clients straight: the skills are real (this page is the proof), the exact products and services are still taking shape, and we'd rather find out with them. Not vendor and buyer: partners, figuring out together what AI is worth in their operation. That honesty opens more doors than any polished pitch.
Pricing goes hybrid: retainer + consumption + outcome bonus, index-linked, 12-month renewable, never 3-year fixed. The market moved the same way: hybrid is now the dominant AI pricing model, and per-seat keeps shrinking. Outcome fees only where we can cleanly measure and attribute them, invoices auto-processed, days off the close. Per-seat is a ceiling anyway: €15 a seat across every user at our ~110 clients tops out around €450k a year, and that assumes everyone buys, which they won't. About 400 seats already subscribe to 360AI today. So we rotate toward outcomes.
Team: a dense bench of managers-of-one
360 org
The FDE ramp runs from 1-2 toward ~10. No juniors, nobody who needs a boss. Each FDE paired with a domain expert, the soft-skill half. The junior maths has flipped anyway: as clients' own AI eats the easy tier, a junior turns from day-one margin into a long investment.
The first FDE hire comes from outside. The most retrainable consultants are also the highest poach risk, so we plant external genetic material first and grow the role internally after. And the final filter on any shortlist is the wiring assessment (CoreTalents), not the CV. 360 already has a trained assessor in house.
We don't recruit on money. Comp is table stakes at 360, not the pitch. And the team's ownership instrument, whether that is equity, phantom shares, royalty or a partner track, gets decided before the first offer goes out, because retention is the game-over variable.
The shape of 360 changes
Today 360 has a sharp focus: Odoo, and only the implementation side of it. That focus served us well. This plan changes it in both directions, on purpose.
Wider. Headless websites and portals, built beside Odoo rather than on it. And the administratiekantoor, a firm we run rather than software we hand over.
Deeper. With FDE work we move into the client's production database and stay there, operating workflows instead of delivering them and walking away.
An implementation partner implements and leaves. The firm this plan describes builds, runs and stays. That is a different company, and it is worth saying out loud: we are choosing to become it.
What we're honest about
Naming these is what makes this a plan and not a pitch. Each one comes with what we do about it.
Risk 1
No proven outcome-buyer. Clients love the demo, the final switch is hard, and nobody is queuing to invest yet.
The cheap test
The production-data POC in August: a working agent on a prospect's own data, before any contract. It sets the pace of everything downstream.
Risk 2
The kerkhof aan apps. Too many fronts, none deep. And the length of this plan is itself the risk.
The counter
One flagship, deep. Width only for clients on Odoo, and billed from day one. A new build idea has to displace something, not add a front.
Risk 3
The load-bearing numbers are back-of-envelope. And written hours smooth over the real change, confirmed firsthand by multiple people in the team.
The fix
Stand up the efficiency baseline nobody owns yet, starting from the budget-vs-actual data we already have, before we price anything on it.
Risk 4
Odoo SA: friend or threat. The squeeze in F-04 is fact, the vendor is climbing the stack. What stays open is how hard that lands on partners, and how fast.
The read
The September reveal is the tell. The end-Aug hypothesis one-pager (3 scenarios) pre-commits our response, so September is execution, not scramble.
Risk 5
Not everyone will make it, and pretending otherwise would be the pitch version. Everyone gets the tools, the time and the support, and still the gap between people who adopt and people who don't will widen. 10x people and 1x people don't work well together for long. Colleagues who can't or won't adopt will have a hard time, here and anywhere in this trade.
The deal
The plan's half: make adopting as easy as it can be made. The repos, the plugin, the first 30 minutes, help always on tap, and no gap ever used against anyone. Your half: take the reps. The support is unconditional. The outcome isn't.
The guardrails
Posture first: support role, boost-don't-replace, no AI-transformation talk. Visible support wins earn the standing. Big-vision talk comes later. And the don'ts, non-negotiable:
No second ERP platformNo selling autonomous finance in 2026No replace-the-GL buildsNo pricing against FTEs-displacedNo 3-year fixed outcome contractsNo rebuilding commodity plumbingNo spreading across verticals before owning oneNo width work without an Odoo core
What this needs from leadership
A plan without an ask isn't finished. Four things, none of them big:
Ask 1
The window stays protected: the two months of change-PM focus already granted, kept clear until September.
Why
The whole P0-P2 sequence assumes it. Everything on the board is sized against it.
Ask 2
Financial and timesheet access. Already on the board, and the one item that needs a push from leadership.
Why
The efficiency case, the migration pricing and the margin numbers all run on estimates until this lands.
Ask 3
The ownership-instrument decision before the first FDE offer goes out, plus a budget envelope for those first 1-2 hires. And not just for new hires: the people already carrying this work count first.
Why
Retention is the game-over variable, and comp isn't the lever. The cheapest attack on this whole plan is one offer letter to one of our own pioneers.
Ask 4
One warm prospect intro and a slice of selling time for the production-data POC. We build the agent, the door is yours.
Why
The POC is the cheap test that sets the pace of everything downstream, and it needs a prospect to run on.
04
Sequenced against fixed clocks
The roadmap, against the clock
Two views of one plan. View one cuts by time: six phases from here to the first outcome contract. View two cuts by work: five streams that run through those phases. Same facts, two lenses.
View 1: six phases, four fixed clocks
The outside clocks don't move. Leadership is back around 1 September. Odoo 20 gets revealed at Odoo Experience, 24-26 September. The 100-day review lands around end of October. And the destination has a date: one customer live on an outcome SLA by end of Q1 2027. Everything internal is sequenced to land against those four.
~1 Sep · leadership back
24-26 Sep · Odoo 20 reveal
~end Oct · 100-day review
Q1 2027 · first outcome
T-0 today
P0
Contouren
now → ~18 Jul
Finish and hand over the plan. This site is the handover, so P0 closes as you read.
Done when · 1
Plan shared, first async spar open
P1
Boost + seed
~14 Jul → early Aug
Visible wins plus the connective glue.
Done when · 6
The version-delta review harness ships. The first visible boost-win
The scoped-key MCP shipped
AI-usage baseline survey live
Three-repo 360 brain + intake skill seeded
Flagship picked
Gijs moving into the R&D seat, his strongest one
P2
Deepen + decide
August
Flagship slice-0 · Odoo 20 prep · the demand test.
One production-data POC run. Does demand show up once we show them?
Odoo 20 hypothesis on one page: three scenarios, response pre-committed
Finance / vertical commit decision
End-Aug gate
P3
Reveal + commit
September
Here's the chapter we connected. Odoo 20 lands, the pre-committed response runs, the finance commit goes to leadership.
Done when · 2
The chapter coheres: brain used, friction visibly down
Odoo 20 friend-or-threat tell answered, same-week strategy update and commit
Leadership back ~1 Sep · Odoo 20 reveal
P4
Prove
Q4 2026 · ~end Oct
The proving quarter.
The 100-day check · 5
1-2 FDE hires in process
One pilot client with an outline outcome contract
Cost benchmark on ≥1 workflow we already deliver
Per-seat → outcome transition outlined
Public AI posture started. Downstream of real capability, never ahead of it
100-day review ~end Oct
P5
First outcome
Q1 2027 (end)
Re-paper one customer to an outcome SLA.
Done when · 1
One existing per-seat customer live on a hybrid outcome SLA: retainer + consumption + outcome bonus
Q1 2027 · the first outcome
Every phase closes on a gate, not a feeling. The done-when lists are the gates. Two deserve saying out loud. P2 ends on the end-Aug gate. And P4 is the proving quarter: if hires, a pilot client and a cost benchmark haven't moved by the 100-day review, the bet is off track, whatever gets said in meetings.
Fork one · the Odoo 20 tell (P3)
Open agent-runtime plus marketplace: ride the wave, build on top. Closed, or Odoo Operate (an Odoo-hosted agent runtime): speed up the substrate-agnostic build, our own stack that doesn't depend on Odoo's runtime. Open-but-unsafe, the current lean: the safe-MCP window is our wedge. The end-Aug hypothesis picks the response up front, so September is execution, not scramble.
Fork two · the demand reality (P2)
If the production-data POC confirms the doubt named in the risks (demo-love without a buyer), internal value and fixed-price arbitrage carry us longer before outcome pricing, and P4-P5 slow down. If real demand surfaces, they speed up. One cheap test sets the pace of everything downstream.
The value curve
Month one is July, the mandate start. Listening and seeding first, nothing visible yet. That's the plan working, not stalling. The hours-saved floor starts covering cost around month three to four, early in Q4.
View 2: five workstreams
The strategy's four buckets are the framing. These five streams are the same work, cut for execution, each tagged with the focus area it serves.
The install base behind the one stream with a date on it
84 / 95of our clients sit on Odoo 16, 17 or 18, and all of them want 20. The 36 on 16 and 17 fall outside Odoo's three-version support window once 20 ships. Each square is one client.
7 on 16 · out of support29 on 17 · out once 20 ships48 on 18 · all want 2011 on 19 · the ones we didn't skip
360AI product · the finance flagship
client
Now
Pick the flagship. Our call: finance, so close, AR, AP. The all-in finance package on paper. Package what exists, so a client question isn't always a live demo.
Next
Slice-0 (the smallest working end-to-end version) at one friendly, high-volume, rules-based client: draft-create plus an approve inbox, an audit gate in front of every write.
Later
Rotate the flagship off per-seat toward outcome pricing, before Odoo 20 flattens the surface underneath it. The horizon: the AI-first administratiekantoor on Odoo.
AI-driven migrations
client
Now
Start the 18→20 delta analysis. The code pins (Odoo freezes the 20 codebase) in a month or two. Triage the 84 by urgency and custom depth.
Next
The migration build week once 20 pins. First 2-3 pilot migrations off the 16/17 list. Decide fixed-price vs hours before the first quote, and pre-sell the wave to the 36 before anyone else quotes them.
Later
Run the 2027 wave, 30 to 40 projects, through one pipeline. The learning loop compounds: an issue seen once is known forever.
FDE & team ramp
360 org
Now
Profile and funnel ready. Manager-of-one is the hard gate. Comp is table stakes at 360, never the pitch.
Next
First 1-2 FDE hires in process, with the ownership instrument decided before the first offer goes out.
Later
The bench toward ~10, every FDE paired with a domain expert. Headcount is the residual, not the target.
Org & the 360 brain
individual360 org
Now
Seed the three repos: playground (no bar to contribute), skills (curated, proven), plugin (one command, every proven flow in your terminal). AI-usage survey live. Start the efficiency baseline nobody owns today. And keep moving the org's own stack to European providers, piece by piece.
Next
The consented intake skill so contributing isn't manual. Onboarding skill and the client-summary pull once the repos carry real weight.
Later
One shared way of working: a new hire runs the plugin and works our way on day one, without a course. Switch-ready the day a credible European frontier model exists. And the brain itself goes on sale: set up your company brain, with us as the working reference.
Clients & pricing
client
Now
One production-data POC: a working agent on a prospect's own data, before any contract, under NDA + DPA and sandboxed. One client AI-architecture session. And width projects (an app, a portal, a headless site) whenever a client on Odoo asks: reactive by design, billed from day one.
Next
A pilot client with an outline outcome contract, plus a real cost-per-outcome number on one workflow we already deliver.
Later
First hybrid outcome SLA live: retainer + consumption + outcome bonus, 12-month renewable, never 3-year fixed.
Why migrations is the stream with a date on it. We steered most clients past 19, so the install base is stacked up behind us. And it's the most AI-tractable work we have: both versions' full source is public, unit tests plus data reconciles make the result checkable, and the unattended 17→19 run in Proof is the receipt. The product is not the code transformation. Odoo will commoditise that. The product is the learning loop tuned on our own 84 and on the EU statutory depth F-07 calls inventory, plus the fixed-price wrapper around it. Every issue hit once is known at the next migration, so migration 30 costs a fraction of migration 3. At the ~€10k a manual migration cost last time, 30-40 projects is a serious revenue year, and the pipeline is what keeps margin in it while every other partner discounts. The premium window is maybe 12-24 months before the tooling commoditises, which is one more reason the wave is an on-ramp to a lasting AI relationship, not the destination. And this only works with our migration expertise in the middle: the build week runs with the people who carry it, not around them. The pipeline takes the repetition. The deep-custom calls, the review authority over everything the pipeline produces, and the OCA work stay theirs.
Why the 360 brain is the foundation, and quietly a product. The brain is how individual wins stop evaporating: every solved problem becomes a skill anyone can run, every migration issue a lesson the pipeline keeps, every new hire productive on day one. It is the compounding engine under all three focus areas, which is why it ships first. The pattern is going mainstream too: Cerebras just published how they built theirs. Three months after launch it answers 15,000 questions a day, from people, automations and agents alike, and practitioners called it what it is, ontology, knowledge base, company brain, all the same thing. And here is the part that matters most: our clients will need exactly this. A curated skills repo, a consented intake, one command to put it in every terminal. We build it for ourselves first, and the moment it carries our thirty people, "set up your company brain" becomes a service with a working reference: us. Bet 6, live.
05
Q3 now · Q4 2026 · 2027
The work, by quarter
Sixteen things in Q3 plus one for everyone, read as a priority order, not seventeen promises. Two are hard commitments: the first client invoice inbox live on 1 August, and the Odoo 20 hypothesis by end of August. Every other window is a target that slips if the flagship call or the Odoo 20 response eats the month, and it should: the priority order doubles as the slip order, bottom first. Overcommitment is the failure mode to watch. Q4 is the proving quarter, 2027 is where it pays back. Those are goals that get concrete as they get closer, and the open migration items get named owners in the September session. Every action has a window and a size. See one you want? Ping Douwe and it's yours.
Showing 30 / 30
Q3 2026 · now to end September
0 / 17
Your first 30 minutes (yes, you)
As soon as the repos are seeded: install the 360 plugin, run one proven flow (Gijs's Docker Fixer is the first), and drop one thing you tried, however half-baked, in the playground. That's the whole ask. No course, no gap to close, just a first rep.
everyonebrainfrom next week · 30 min
Seed the three repos
Playground open to everyone from day one. Skills starts with one proven flow in it (Gijs's Docker Fixer), not an empty structure nobody fills. The plugin puts every proven flow in a consultant's terminal with one command.
Douwebrainthis week · weeks
Version-delta review harness
An 18→19 environment reviewed for under €5 a run. Cheapest visible win we have, and every ingredient is already owned.
Douwebrainthis week · days
Scoped-key Odoo MCP
The first scoped, whitelisted key ships to the builder who asked for it. Oldest open promise on the list, from 6 July. More keys follow right after. It unblocks the chapter's MCP work.
Douwemcpthis week · hours
First client invoice inbox live
One inbox, AI routes every invoice to the right entity. Two of the four entities first, reusing a flow we already built elsewhere, staging before go-live on 1 August. Our first AI flow in a client's production, which beats any demo.
committedin motionGijs + Douwedemand1 Aug · days
AI-usage baseline survey
A couple of colleagues sanity-check the draft first. Framed as show us what works, results published back within the week. Measure to boost people, never to audit them.
DouwemeasureJul · a week
The all-in finance package on paper
The all-in finance idea has been carried in the team for a year, and the flagship call finally gives it a lane. On paper first, then a working session to land it, including what it needs this time that it didn't get before.
open · claim itflagshipAug · hours
Start the 18→20 delta analysis
Code pins in a month or two, so this starts now. OCA's OpenUpgrade for 20 trails by about half a year and our self-hosted base needs its own tooling regardless. Odoo's own Odev AI aims at this same work, but at odoo.sh, not our self-hosted base: it sets the ceiling on migration pricing, not the floor. What we learn flows back upstream to OpenUpgrade where it fits.
Douwemigrationnow → Sep · weeks
Triage the 84 by urgency and custom depth
Which of the 36 on 16 and 17 go first, which are deep-custom and slow, which are near-standard and can prove the pipeline. Input to both the build week and the pricing call.
open · claim itmigrationAug · days
Check if Odoo 20's MCP ships OAuth
Gijs lifted his 19 MCP scope from the 20 codebase but couldn't verify this one. It sets how wide the safe-MCP window is. Hours of work for a bet-sized answer.
Gijs + DouwemcpJul · hours
Odoo 20 hypothesis, one page
Three scenarios and the response we commit to, written before the reveal. That way September is execution, not scramble.
committedDouweorg designend Aug · days
Instrument the efficiency number
Start from the budget-vs-actual milestone data we already have, not from zero. Is the work genuinely faster, or do more tickets just come in? Nobody owns this number today, and the baseline rots while we wait. The full baseline follows the access ask, so the pricing calls wait for it.
DouwemeasureAug · weeks
The intake skill
Consented auto-harvest of prompts and flows into the repos, preview before send, never client data. Manual knowledge bases die on the contribution problem. This is the fix.
DouwebrainAug · weeks
One client AI-architecture session
Gijs and Douwe plus a senior colleague at a receptive client. What's possible across their whole org, not a 360AI demo. The client asked us for exactly this after seeing the first flows, so the demand is already sitting there.
Gijs + DouwedemandAug-Sep · hours
Gijs into the R&D seat
Gijs moves to where he's strongest: 0-to-1 R&D and the creative work. He proposed the split himself. Douwe picks up the client side and the internal connecting work. Agreed all round before it's official, and official before September.
in motionDouwe + Gijsteambefore Sep · weeks
Financial and timesheet access
The efficiency case, the migration pricing and the margin numbers all run on estimates until this lands. The one item on the page that needs a push from leadership.
DouwemeasureQ3 · hours
The stack goes EU-first, piece by piece
360's own tooling moves to European providers, one piece at a time. Already underway. Claude and Google remain until a credible European frontier model exists, and the day it does we can switch, because everything around it already lives here.
in motionDouweorg designcontinuous
Q4 2026 · the proving quarter
0 / 8
The migration build week
Lock two or three people away for a week, come out with the scripts and the agent, trained on 19 and 20 source. Runs once the 20 code pins, with our migration specialists as the design centre: they call what the pipeline may touch, and review authority stays with them. Owners named in the September session.
open · claim itmigrationOct · a week
First pilot migrations through the pipeline
Two or three, starting with near-standard clients off the 16 and 17 list. They prove the loop and they have a support deadline anyway. Every issue they hit feeds back into the agent.
open · claim itmigrationQ4 · the quarter
Decide migration pricing
Fixed price or hours. The cleanest fixed-price-arbitrage candidate we have: repeatable, scriptable, bounded. Decide before the wave, because after the first quote it's set.
open · claim itmigrationQ4 · a decision
Flagship slice-0 building
One friendly existing client, high-volume and rules-based, with a clean outcome metric. Draft-create and an approve inbox first. No agent writes anything without the audit gate in front of it.
DouweflagshipQ4 · the quarter
First FDE (forward-deployed engineer) hires in process
One or two, with the ownership instrument decided before the first offer goes out. Comp is table stakes at 360, so it isn't the pitch and it isn't the retention lever either.
Douweteam~end Oct · weeks
One pilot client and a cost benchmark
A willing client with an outline outcome contract, plus a real cost-per-outcome number on one workflow we already deliver. This is what the 100-day review checks.
Douwedemand~end Oct · weeks
Onboarding skill and the client-summary pull
Both wait for the repos to carry real weight. The client-summary pull is the one people keep asking for unprompted: what does this client do, what's running, what are their flows.
open · claim itbrainonce the repos hold · weeks
First width project, when a client asks
An app, a portal or a headless site for a client on Odoo, connected to their Odoo. Reactive by design: taken on the moment it's asked for, billed from day one, delivered the way the proof band shows we can.
open · claim itdemandwhen a client asks
2027 · where it pays back
0 / 5
Run the migration wave
30 to 40 projects on Gijs's estimate, and the install base says that's the floor. The year the tooling pays for itself many times over, or the year we find out we were late.
open · claim itmigration2027 · the year
First outcome contract live
One existing per-seat customer re-papered to a hybrid outcome SLA: retainer plus consumption plus an outcome bonus. Never pure-outcome, never 3-year fixed.
open · claim itdemandQ1 2027 · weeks
The migration agent matures
An issue seen once is known forever, so migration 30 costs a fraction of migration 3. The compounding asset, and it only exists if we design it in from the start.
open · claim itmigrationthrough the wave
The FDE bench toward ten
Every hire a manager of one, each paired with a domain expert. Headcount is the residual, not the target. We speak in revenue-per-FDE and agent reuse instead.
open · claim itteam2027 · the year
Rotate the flagship off per-seat
Per-seat caps out. The ceiling arithmetic lives in the strategy. Move the flagship to outcome pricing before Odoo 20 flattens the surface.
open · claim itflagship2027 · the year
Ticks you make live in your own browser only, so they won't show for anyone else. To claim an action or mark one done for the whole team, ping Douwe: he writes the shared state into the site everyone opens. Reset clears only your own ticks.
06
Where this ends up
The north star, restated
The leading AI-native Odoo partner in the EU. Not a bigger 360: deeper into the client, wider around their Odoo.
And at the far end of the arc: the first AI-first administratiekantoor on Odoo. Ours.
This site is itself a proof point. It was made the way the plan says work gets made now. Dedicated agents drafted each section from the research wikis, and five sweep agents checked every claim against the sources, one per content section. Three design agents pitched competing directions, a market scan ran outside-in, and review agents read the result through five lenses: the partners, the team, an AI-savvy colleague, a client, and Douwe himself. Every segment then got a blind review from an agent that saw nothing else, so each part has to stand on its own. Then a human touched it up. Boost, don't replace: the judgment stayed human, the repetition didn't. (Douwe mostly asked nicely.)
It's also a living document. When the flagship call lands, the Odoo 20 tell arrives, the POC answers the demand question, or the first width project comes in, the site changes with them. Argue with it, that's what it's for: hit the Argue button bottom-right, or ping Douwe.
Sections drafted by5 dedicated agents
Claims checked by5 sweep agents
Design directions3 pitched · 1 chosen
Review lensespartners · team · AI-savvy · client · Douwe
Blind segment reviews7 · one per segment, in isolation
Sources2 research wikis
Human in the loopalways
360, AI-native · July 2026 · internal · living document
