Xerocon Denver 2026 wrapped last week, and every recap you’ve read since was written for firms.
This one is for you: the founder with a Xero integration, a marketplace listing, a roadmap meeting on Thursday, and a board asking what the announcements mean for the plan.
2026 was the year both platforms told you, in public and at length, which parts of your product they intend to build themselves. Below is what they said, and the five places worth building instead.
If you only read one line: build vertical depth, the exception layer, jurisdiction-deep compliance, systems of action and an agent surface, and stop building anything the platforms shipped natively this year.
What Xero and Intuit changed in 2026
Three moves, run in the same order by Xero and Intuit.
1. The API stopped being free. Xero announced it in December and landed it in March. App Store subscription billing is retired, replaced by tiers priced on connections and data egress: nothing at Starter, around $895 a month at Advanced, egress allowances stepping 10GB, 50GB, 250GB, overage by the gigabyte, and partners asked to move customers onto their own billing by 30 June. Intuit got there first and framed it more gently. Core calls (mostly data-in) are unmetered, CorePlus calls (mostly data-out) burn monthly credits, and tiers run from free to $4,500 a month.
Xero used to charge you when it introduced you to a customer. Now it charges you for the data every customer consumes, including the ones you brought yourself.
2. The middle of the stack got absorbed. Category by category, with both platforms’ 2026 answers side by side:
- Document capture. Smart Document Capture, pitched explicitly against traditional OCR tools. Intuit’s Accounting Agent categorises, resolves anomalies and reconciles.
- Bill payment. Xero Bill Payments, on top of a $2.5bn Melio acquisition. Intuit has done this for years and now has an agent that does the chasing.
- Expense management. Melio Expense Management tracks card spend in real time, and Xero made a point of the fact that customers don’t have to change cards. That’s a shot at the spend management category’s toll booth.
- Payroll. Xero Payroll powered by Gusto closes the biggest hole in the US line-up.
- Close management. XeroForce landed in Denver with a month-end agent that works through document status, reconciliation status and prepayment and amortisation journals, then hands its working back for review. Intuit prices the same job per client per month.
- Client chasing. Casper identifies missing information and contacts clients through the month. Intuit’s redesigned Inbox does it with SMS nudges.
- Collections and AP risk. Payment Follow Ups, Bill Protection and Cash Flow Actions on one side; Customer and Finance agents on the other.
- Reporting and consolidation. Xero Ultra bundles Syft consolidation and advanced reporting into a mid-market tier. Intuit Enterprise Suite does consolidated entities and live KPIs.
- Practice operations. Partner Hub picks up portfolio-wide book health, job status and month-end readiness from October, with JAX chasing the transactions missing a receipt. Intuit’s Accountant Suite replaces QuickBooks Online Accountant at the end of the year, with Workflow Automation in beta and an Agent Studio in preview.
Every one of those categories supported standalone businesses in 2024. All of them now ship inside the ledger, with a platform’s marketing budget behind them.
3. Distribution moved from the listing to the agent. Xero’s MCP server usage grew ten-fold between December and May, past a million API calls by June. New app registrations are up four-fold. One in five connections into the Xero ecosystem is now a custom app that never went near the App Store, which is why there’s suddenly a Development Partner of the Year award. Intuit’s MCP server publishes 144 tools, its surfaces have been live inside ChatGPT since March, and February’s Anthropic partnership is built so a business can run an agent across its ledger data and the data in its own apps.
First, which market are you in: SMB apps or practice apps?
Most of the commentary this month has treated the ecosystem as a single audience, when the same announcements land very differently depending on who signs your invoice.
If you sell apps to small businesses
Almost everything absorbed in 2026 sits on your side of the ledger: capture, bill pay, card spend, payroll, collections, cash flow planning, and consolidation for the clients outgrowing standard Xero. Your competitor is now bundled, free at the point of use, and one click from a screen your buyer already has open.
Two things follow. Your product has to survive the sentence “but Xero does that now”, and the only answers that hold are depth in a vertical, depth in a jurisdiction, or the messy work the native version won’t touch. And your API bill scales with how much data each customer pulls, so egress per customer is the number to watch.
If you sell apps to accounting practices
This is the shift the ecosystem has under-reacted to. 2026 was the year both platforms walked into practice operations.
Partner Hub. From October, firms get live book health, job status and month-end readiness across the whole portfolio in a single view, with JAX chasing the transactions that are missing a receipt. That is the practice dashboard, free, inside the product every firm already opens each morning.
XeroForce. A no-code agent builder, and from Denver a pre-built month-end agent that works through document and reconciliation status, posts prepayment and amortisation journals, and hands its working back for a human to accept. Look past the demo to the shape of it. Xero never has to build the long tail of small practice workflows. It ships the builder and a starter template library, and lets firms and partners build them.
Intuit is running the same play. Accountant Suite replaces QuickBooks Online Accountant at the end of the year, Workflow Automation is in beta, Agent Studio is in preview, the Inbox centralises client requests, and cross-client portfolio insights arrived in August.
Three things survive that:
- The work the ledger can’t see. Jobs, capacity, WIP, billing, engagement letters, staff time, lodgement and filing status, the compliance calendar. Partner Hub reports on ledgers. A practice runs on everything wrapped around them.
- Every client who isn’t on Xero. Partner Hub will only ever show the Xero half of a portfolio, and Accountant Suite the QuickBooks half. A practice tool that covers Xero, QuickBooks, MYOB and Sage is describing the firm’s actual client list, which is the one view neither platform has any reason to build.
- Being the firm’s system of record, with agents calling into it. Theirs and yours. That is also the product the firm’s buying committee can actually say yes to, because it survives a change of ledger.
One piece of arithmetic before you plan the year. Xero’s new tiers cap connections as well as data, from five at Starter to ten thousand at Advanced. An SMB app counts one connection per customer. A practice app counts one per client file, so 300 firms averaging 200 clients each is 60,000 organisations. Work out which tier that puts you in before the renewal lands.
The 5 places worth building in the Xero ecosystem
They apply on both sides of that split, and they get harder to absorb as you go down the list.
1. Verticals the ledger will never learn
Look at who won this year’s App Awards: rostering and invoicing for disability and aged care, inventory and jobs, marketplace settlement reconciliation.
General ledgers generalise. They will not learn NDIS line items, construction retentions, trust accounting rules or per-marketplace settlement quirks, because each one is too small a market for a platform and plenty big for a founder. Depth in a vertical is the one moat that gets wider as the platform gets more general.
Quick test: could a Xero PM describe your product to their own leadership in one sentence, without using a single industry-specific noun? If they could, you’re in the absorption zone.
2. The exception layer, with its working shown
Every demo this year ran on clean data. Every firm you sell to lives in the 4% that breaks: the Stripe payout covering forty invoices and three fee lines, the supplier who changed bank details, the journal that genuinely needs a human.
Native automation will take the happy path, and it will be free or close to it. What stays paid is the product that handles the mess, shows its working, and produces a reviewable trail a partner can sign off. An agent that files journals unsupervised is a liability. An agent that drafts them and shows how it got there is a junior who never gets tired.
Build for the exception and you’re selling the thing the platform can’t ship at version one.
3. Compliance that’s jurisdiction-deep
Payday super, AML tranche two, MTD, e-invoicing mandates, US sales tax nexus.
Regulation moves faster than platform roadmaps and is region-locked by definition, which is exactly why Xero’s own awards went regional this year. It’s the least fashionable moat in the ecosystem and the most durable one, because the deadline does your marketing for you and no platform can prioritise eight jurisdictions at once.
Compliance depth travels badly between markets, though. Read the real differences between the UK, AU and US markets before you assume one build covers three.
4. Systems of action
Intuit leaves data-in unmetered and charges for data-out. Xero charges egress by the gigabyte. Between them they’ve published, in effect, a price list for lazy architecture.
The economics have flipped toward products that put work into the ledger and away from products whose main activity is pulling data out to re-display it. If your product is a nicer view of numbers that already exist, your differentiation and your unit economics are under pressure from the same direction, and Ultra and Enterprise Suite are coming for the top of that market anyway.
Practical version: kill polling, move to eventing and webhooks (Xero’s Horizon early adopters reported close to half their GET calls disappearing), and be able to state gross margin per customer including API cost. If you can’t produce that number today, that’s this quarter’s work.
5. Being callable by other people’s agents
Don’t bolt a chat box onto your UI. Ship your product as tools another agent can call: clear names and scopes, permissions a firm can reason about, deterministic actions where determinism matters, and a log of everything the agent did inside your product.
We’ve made the case before that a marketplace listing converts existing demand. The 2026 update is that a growing share of buyers never walk down the aisle at all. They ask an agent, or they build one. With one in five connections now custom-built, there’s a second business hiding in here: the connectors, templates, agent components and monitoring that firms building their own tooling are currently improvising.
Good automation is predictable. A firm should be able to say what your product will do before it does it, every time, which is the difference between a system and a slot machine. The vendors who get that right are the ones agents keep choosing.
The hedge that sits across all five: your second ledger
Casper works on every ledger, not just Xero. Of everything announced in Denver, that detail carries the clearest instruction for everyone else.
Single-platform apps carry single-platform risk, from absorption and from a unilateral pricing change you don’t get a vote on. Going deep on one platform first is still right; you just need to know in advance which platform is second and what would trigger the move. QuickBooks Online Accountant retires at the end of this year, which makes the next few months the cheapest door into QuickBooks practices you’ll get this decade.
And 5 things to stop building
The same year of announcements, read the other way.
- Standalone OCR. Capture is table stakes inside both ledgers now. Capture wrapped in a workflow the platform doesn’t have still sells; capture on its own has become a feature.
- Generic small business AP. Both platforms pay bills natively, one of them having spent $2.5bn for the privilege.
- Read-only reporting and dashboards. Absorbed at the top by Ultra and Enterprise Suite, at the practice end by Partner Hub from October, and squeezed at the bottom by egress fees for the privilege of fetching the data.
- Chat-with-your-books wrappers. Native on both platforms, and both are already inside Claude, ChatGPT and Microsoft 365. There’s no wrapper margin left.
- Client chasing as an entire product. Casper on one side, the Intuit Inbox on the other. Chasing is one feature inside a larger product now, and the larger product holds the pricing power.
What this does to your go-to-market
Your positioning now has to answer a roadmap. “Better than the incumbent app” worked in 2024. The question in the room today is “why won’t Xero just do this?” You need a real answer, built from depth, jurisdiction, exceptions or coverage, and it belongs in the first line of your website, long before the fourth slide of the deck.
Integration depth is your defensibility and your marketing. An integration that becomes part of the firm’s workflow is a control point. A thin one is a rounding error on someone else’s roadmap. Depth is what earns the partner-team relationships and the conference presence that actually drive volume.
Per-seat pricing assumes flat cost to serve. That assumption died in March. Outcome pricing (per entity closed, per return filed, per reconciliation completed, per exception resolved) aligns what you charge with what you cost, which is the only shape that survives a metered API. Start from pricing models; a percentage increase on the current sheet won’t cover it.
A sensible order of operations
- Instrument API usage per customer this month. Both platforms are billing on it, and you can’t price what you haven’t measured.
- Kill polling. On Xero’s own numbers that’s roughly half your GET calls gone.
- Write down what you do that the platform won’t do within twelve months. If the honest answer is a feature list, fix that before anything else on this list.
- Ship an agent surface. Tools, scopes, audit trail, docs a model can read.
- Re-run pricing against your new cost to serve, and move toward outcomes.
- Name your second ledger, or write down why you don’t need one.
- Then reopen the roadmap you had before Denver and cut everything that just became a feature.
Both companies spent 2026 saying out loud what they intend to build, across two conferences and a pricing note. The expensive mistake now is planning your roadmap as though they hadn’t.
Working out where your product sits against the platform roadmap is go-to-market strategy work, and it’s most of what we do. Related reading: getting found in the app stores, control points, and pricing models that survive contact with the market.
