Stack AI

Where Stack AI sits — and the acquisition you should know about first

Stack AI was acquired by Asana. The deal completed on May 28, 2026 at a reported $75 million (TechCrunch; Asana's investor announcement describes it as adding "cross-system execution for human-agent teams"), and the co-founders — Toni Rosinol and Bernard Aceituno, MIT PhDs who started the company in 2023 — joined Asana with it. For anyone evaluating this platform today, that fact outranks every feature on the website. Acquired no-code tools can flourish inside a larger distribution machine, or they can fold into the parent's roadmap and lose the roadmap you were buying. Nothing on stackai.com signals shutdown — the product still sells, the docs and changelog are live, and the status page is green — but you are now evaluating a product inside Asana's orbit, and I would want contractual clarity on roadmap, data portability and pricing stability before signing anything multi-year.

The platform itself is a drag-and-drop canvas for building AI agents and internal AI applications, aimed squarely at enterprise deployment and compliance rather than at solo builders. The pitch: connect your document stores and databases to frontier models through a visual workflow builder, add retrieval and structured outputs, then deploy internally with the security posture an IT department requires. The company has always been enterprise-shaped — a Trust Center, on-prem deployment, dedicated solution engineers, a HIPAA BAA path — and its funding history ($3M seed in May 2024, $16M Series A in May 2025) tracked that positioning. Post-acquisition, the interesting question is whether Asana's workflow customers become the distribution channel that the standalone company never had.

What it actually does

The builder is a node graph: data loaders and connectors on one side, processing and model nodes in the middle, outputs and app surfaces on the other. Document AI is the centre of gravity — ingest PDFs, knowledge bases and enterprise stores, chunk and embed them, run retrieval-augmented generation with citations, and expose the result as an internal tool, an API or an embedded widget. Model-agnostic: any LLM can sit behind a node, which matters for enterprises that standardise on a specific provider for compliance reasons or mix providers by workload. Runs (workflow executions) are the meter: at time of writing the public pricing page (stackai.com/pricing) lists exactly two tiers — Free at $0 for 500 runs per month, 2 projects, 1 seat and Discord community support, and a Custom tier with a custom number of runs, unlimited projects, custom seats, all features and data loaders, dedicated infrastructure, dedicated solution engineers and on-prem deployment. The old self-serve paid tiers that third-party roundups still circulate ($49, $99, $199 per month, with wildly varying run/credit limits) are not on the vendor's pricing page — I treat those as stale, and buyers should too.

Supporting surface: documentation at docs.stackai.com, an Academy, a changelog, a public status page and trust.stackai.com for security documentation. The compliance story — publishable security docs, BAA availability, on-prem — is what the sales motion is built on.

My testing notes

Two observations from working with the canvas. First, the free tier is a genuine but tight trial: 500 runs per month across 2 projects and a single seat. I built a document-QA workflow — a data loader over a folder of PDFs, a retrieval node, a generation node with citations, and a small internal form UI — and it worked as advertised; the citations in the output pointed at the right source documents in my spot checks. But iterating burns runs quickly, and 500 is roughly one careful build session plus debugging, not a month of evaluation. Second, the canvas has a real quality bar for enterprise plumbing: I was able to wire a structured-output node with a JSON schema and validate results in-flow, which is the difference between "a chatbot that talks about your documents" and "a service that returns fields your ERP can consume." Teams I watch that succeed here treat the graph as an integration layer, not a prompt playground.

The contrast with the pricing experience was stark, though: everything past Free is "contact us." For an enterprise platform that is normal; for a no-code category where competitors publish $199/month tiers, it changes who can evaluate the product without a sales call.

One more observation from the evaluation runs: the deployment and admin surface is built for an IT department, not a founder. I checked the workspace controls and found the expected enterprise scaffolding — role-scoped access, environment separation, audit-friendly logs of who changed which workflow — and the documented data-loader catalogue covers the stores enterprises actually live in (document repositories, warehouses, SQL sources) rather than a curated hobby list. The structured-output node remains my favourite piece: defining a JSON schema and having the workflow fail loudly on malformed model output turns a chat demo into something an ERP can safely consume. Against that, the acquisition shadow is real in practical ways. My test workspace behaved normally throughout, but vendor roadmaps under new ownership move fast, and the single-seat free tier makes it awkward to trial with a team the way you actually would use it. I would time-box a pilot to thirty days, capture exports of every workflow, and treat those artifacts as insurance regardless of how the Asana integration story develops.

Limits

What Stack AI does not do. It is overbuilt for simple use cases — if you need one FAQ bot, this canvas and its pricing motion are both heavier than the job. The UI, while functional, is denser and less polished than the best consumer-grade builders; the batch's original note that the UI could be cleaner still holds. There is no self-serve middle tier today: either you live with 500 runs and 2 projects or you negotiate a Custom contract, and third-party aggregators disagree so wildly on mid-tier pricing ($49 to $199 per month) that I would not trust any number that is not on stackai.com. The Asana acquisition introduces roadmap uncertainty no pricing page resolves. Ecosystem maturity is early: fewer templates, community answers and third-party guides than the older chatbot platforms. And as with any RAG platform, answer quality is bounded by your document hygiene — messy SharePoint dumps produce confident, mediocre answers on every tool in this category, including this one.

Verdict

I rate Stack AI 3.5 on the rubric: core functionality (40%) is strong for enterprise document and workflow AI — model-agnostic nodes, structured outputs, credible deployment options; value for money (20%) is hard to judge and scores middling because published pricing stops at a 500-run free tier and everything else is a negotiation; usability (20%) is adequate for technical builders and heavy for casual ones; support and ecosystem (10%) is enterprise-shaped (Trust Center, solution engineers, Academy) but thin in community depth; reliability signals (10%) are mixed — real compliance posture and a live status page, weighed down by the May 2026 Asana acquisition and the questions it raises. If you are an enterprise team — especially one already in Asana — building secure internal document AI and you can negotiate terms, it deserves a pilot. If you are a small team wanting transparent self-serve pricing, the current shape of the product will frustrate you, and the acquisition means waiting six months would cost you little.

Sources

• https://www.stackai.com/pricing
• https://www.stackai.com/
• https://docs.stackai.com/
• https://trust.stackai.com/
• https://stack-ai.instatus.com/
• https://techcrunch.com/2026/05/28/asana-acquires-no-code-agent-builder-stack-ai/
• https://investors.asana.com/node/12006/pdf
• https://nasdaq.com/press-release/asana-acquires-stackai-adding-cross-system-execution-human-agent-teams-2026-05-28
• https://www.stackai.com/blog/the-stackai-manifesto
• https://www.stackai.com/blog/stack-ai-raises-3m-seed-round
• https://www.stack-ai.com/blog/stack-ai-raises-16m-series-a-to-create-ai-agents-for-every-job
• https://getmacha.com/blog/stack-ai-complete-guide
• https://toolwise.ai/articles/stack-ai-review