- Instinct, a personal AI agent people text, raised $1 billion at a $10 billion valuation in September 2026, about four times its value a month earlier.
- Chatbots are a good front door: they answer leads instantly. But the work that moves deals is the follow-through, and that's where an AI agent comes in.
- 82% of agents in an RPR survey use AI, but only 48% feel confident using it for client-facing work.
- Keep a human approval step for money, contracts, pricing advice and anything that could raise Fair Housing risk.
- Softpers builds real estate AI agents from $3,000, starting with a 1–2 week fixed-price pilot and going live in 2–8 weeks.
On September 28, 2026, Instinct announced a $1 billion Series C at a $10 billion valuation, with Sequoia Capital, Benchmark and Coatue investing. About a month earlier, founder Noah Shinn had told the Wall Street Journal that the company raised $250 million at a $2.5 billion valuation, as Reuters reported. Instinct sells nothing to real estate. But the reason investors quadrupled its value in a month matters to every brokerage that added a chatbot in the last two years.
What happened
Instinct's valuation went from $2.5 billion to $10 billion in about a month.
A $250 million Series B reported in late August (Reuters, citing the Wall Street Journal), then a $1 billion Series C on September 28, 2026.The round is $1 billion, not $10 billion; the $10 billion is what investors say the company is worth. Instinct is still invite-only and hasn't published revenue. On the Invest Like the Best podcast, Shinn said it is growing roughly 10% day over day without marketing spend, that more than $1 billion a year now flows through the platform, and that he spends 40% of his time thinking about compute.
So treat the valuation as a bet on direction, not proof of a finished market. The direction is the useful part: money is moving from AI that talks to AI that does.
What Instinct actually does
Three design choices stand out:
- No app. It works over text, calls and email, the channels people already use.
- Its own identity. It has its own phone and its own computer rather than borrowing yours.
- Agents that coordinate. A trusted network lets two people's agents coordinate directly, for example to agree a meeting time.
None of that is a better conversation. All of it is follow-through.
Where chatbots stop
Real estate did not miss the AI wave. In a February 2026 Realtors Property Resource survey of 225 agents, 82% said they actively use AI in their business. But only 48% felt confident or very confident using it for client-facing work, and accuracy (63%) and compliance (49%) were their top concerns.
Picture a typical evening. A buyer asks a website chatbot at 9 p.m. whether a listing is still available. The chatbot answers in seconds, which is exactly its job, and captures the lead. Then the lead waits until someone sees it the next morning, by which point the buyer may have booked a showing with another agent. The chatbot did its part. Nothing was set up to do the next part.
The real work in a transaction is coordination
Between an accepted offer and closing, a deal runs on inspection, appraisal and financing deadlines, document collection, and status updates for buyers, sellers, lenders, inspectors, title companies and attorneys. Missing one item can cost the deal. That's the kind of multi-party, follow-up-heavy work Instinct is built for, and it's where a chatbot and an agent work best together.
| Chatbot | Chatbot + AI agent | |
|---|---|---|
| New buyer lead at 9 p.m. | Answers questions, captures contact details | Also qualifies the buyer, books a showing in the agent's calendar, logs it in the CRM and texts a confirmation |
| Inspection deadline in three days | Answers if someone asks | Notices the date, chases the inspector and alerts the agent if the report isn't in by tomorrow |
| Lender asks for missing documents | Not involved | Requests them from the buyer, tracks receipt and confirms with the lender |
| Deal closes | Not involved | Schedules post-close check-ins and the review request |
Four lessons brokerages can take from Instinct
1. Meet clients where they already are. Instinct skipped the app. Buyers and sellers text, so your chatbot and agent should work over text too, not only in a website widget.
2. Give the agent its own identity. A dedicated number, inbox and system accounts for the AI, not an agent's personal login. Everything it does is then easy to audit and easy to switch off.
3. Expect agents to talk to agents. The other side's coordinator, the lender and the title company will increasingly run their own AI. Systems that exchange structured updates, not just emails, will move deals faster.
4. Permissions are the product. Shinn says users take time to trust an agent with their credit card, inbox and calendar. In real estate the stakes are higher: the Realtor Association of Sarasota and Manatee calls Fair Housing the biggest AI risk for agents. Keep a human approval step for money, contract language, pricing advice and which neighborhoods a client is shown. And an AI should never send or change wiring instructions.
How to start without a $1 billion budget
- Pick one workflowSpeed-to-lead or under-contract-to-close. Both are high volume and low judgment.
- List the actions, not the questionsCheck calendar, book showing, update CRM, text confirmation. That list is the agent's job description.
- Mark where a human must say yesEvery step that touches money, contracts or advice needs approval.
- Connect the systems the work touchesYour chatbot, CRM, calendar, showing software and transaction platform. The agent needs to write to them, not just read them.
- Run it in shadow mode firstThe agent drafts and proposes; a person approves and sends. Fix mistakes before it acts alone.
- Measure outcomesTime to first response, showings booked, missed deadlines and hours per transaction, not number of conversations.
How Softpers builds it
- Pilot: 1–2 weeks, fixed priceWe map one workflow with your team, connect it to a copy of your real data and run the agent in shadow mode, so you see what it would do before it does anything.
- Build: live in 2–8 weeksWe connect the agent to your chatbot, CRM, calendar and transaction tools, add approval steps for money, contracts and advice, and give it its own number and inbox. Timing depends mostly on how many systems it touches.
- Run and improveEvery action is logged, so your team can see what the agent did and why. We monitor it, review mistakes with you and widen what it handles one step at a time.
What it costs
| What you get | Time and cost | |
|---|---|---|
| Pilot | One workflow mapped and running in shadow mode on your data | 1–2 weeks, fixed price |
| Production agent | Connected to your systems, with approval steps, its own number and inbox, and logging | From $3,000; live in 2–8 weeks |
| Running costs | AI model usage and hosting | Grows with message and task volume; we estimate it before the build |
Running costs are worth taking seriously. Agents that work in the background use far more computing than a chatbot that only replies when asked; it's why Instinct's founder spends so much of his time on compute. A good build keeps those costs predictable by giving the agent only the work it's good at.
The honest limits
A mistaken booking or a missed contingency costs more than an awkward chatbot answer. Agents need clean data, reliable connections to your systems and someone watching the logs. Some tasks legally require a licensed person; the agent prepares that work, it doesn't replace the license holder. And none of this works if the team doesn't trust it, which is why shadow mode comes first.
Whether Instinct ever earns its $10 billion is an open question. The direction it points to isn't. Your chatbot already answers the question. The next question for a brokerage is what its AI actually finishes.
FAQs
What is the difference between an AI chatbot and an AI agent in real estate?
A chatbot handles the conversation: it answers questions instantly and captures the lead. An AI agent handles the follow-through: it qualifies the buyer, books the showing in the agent's calendar, updates the CRM, sends confirmations and chases deadlines, asking a human to approve anything sensitive.
Should we replace our website chatbot with an AI agent?
No. The chatbot is still the fastest way to answer a buyer at 9 p.m. The agent sits behind it and does what comes next. The strongest setups connect the two, so the chatbot's conversation turns straight into a booked showing and a logged CRM record.
What does an AI agent for a brokerage cost?
Softpers AI agent projects start from $3,000. Each one begins with a 1–2 week fixed-price pilot on a single workflow, and most agents go live in 2–8 weeks depending on how many systems they connect to. Running costs for model usage and hosting depend on volume, and we estimate them before the build.
Which real estate tasks should an AI agent never do on its own?
Anything involving money movement or wiring instructions, contract language, pricing or legal advice, and decisions about which homes or neighborhoods a client is shown, because of Fair Housing risk. The agent can prepare that work, but a licensed person should approve it.
Where should a brokerage start with AI agents?
Pick one high-volume, low-judgment workflow such as speed-to-lead or under-contract-to-close. Map every action, mark the steps that need human approval, connect the CRM and calendar, and run the agent in shadow mode before it acts on its own.
- Viral AI agent Instinct raises $1B Series C at a $10B valuation (TechCrunch, September 28, 2026)
- AI agent firm Instinct raises $1 billion in latest funding round (Reuters, September 28, 2026; cites the Wall Street Journal on the $250 million Series B)
- Noah Shinn – Instinct: The Personal Agent, Invest Like the Best episode 493 (Colossus, September 28, 2026)
- Agents want more (and better) AI tools — and more training (Real Estate News, February 16, 2026, on the RPR survey)
- AI in Real Estate in 2026: The Opportunity Is Huge, But So Is the Liability (Realtor Association of Sarasota and Manatee, ELEVATE, March 19, 2026)
