---
title: Nova just got six new powers, and it never picks the wrong one
description: One agent runs all your agents. It remembers you, counts instead of guessing, works after you close the tab, and cannot spend past your limit.
published: 2026-08-18
category: announcements
source: https://buildyourfuture.app/blog/nova-capability-program
---

You have agents. One handles support, one watches the market, one writes the reports. And somewhere along the
way you became their receptionist. Four tabs open, answers copied between them, and you holding in your head
which one knows what.

**Nova is the agent that takes that job off you.** You ask Nova. Nova asks the others. One answer comes back.

Today Nova gets six new powers: it remembers you, it counts instead of guessing, it keeps working after you
close the tab, it clones itself to move faster, it lets you rewind and take a different path, and it runs on
schedule without being asked.

Nine abilities. One switchboard. Every one a switch you flip.

<Shot
  name="abilities"
  width={3000}
  height={3000}
  alt="Nova's settings dialog listing all nine abilities: System Shell, Recall, Pinboard, Routine, Toolbelt, Workbench, Autopilot, Splitbrain and Rewind, each with a switch, above the model and region it runs on"
/>

## It always knows who to ask

Ask a question, get one reply instead of three. Nova reads the request, picks the agent that can answer it,
and comes back with the answer written for you. Ask something bigger and it splits the job across several
agents at once, then writes the summary itself.

Across five straight releases Nova has picked the right agent **100% of the time**, checked at the receiving
end rather than by asking Nova how it did.

Adding an agent takes one gesture: drag it onto the plane. It is in the mesh.

<BenchBars
  caption="Three things decide whether an orchestrator earns its keep: does it ask the right agent, does it reach for the right ability, and does it go find a tool instead of giving up. Nova's current numbers."
  source="Nova 0.51.0. Three runs per case, temperature zero, on a deliberately mid-range model."
  label="score"
  highlight="asks the right agent"
  points={[
    { x: "asks the right agent", y: 1.0 },
    { x: "picks the right capability", y: 0.932 },
    { x: "finds a tool for the job", y: 0.867 }
  ]}
/>

## Recall: you never repeat yourself again

Tell Nova once that invoices go to your accountant, that the Berlin office shuts in August, that everything
quotes in euros. Next month, in a fresh conversation, on the dashboard or over Telegram, it already knows.

Long conversations stay intact too. Nova folds older turns into a summary as it goes and keeps every original
searchable, so turn two hundred can still answer a question about turn one.

<Shot
  name="recall"
  width={3000}
  height={2000}
  alt="The Recall panel listing durable facts Nova has stored, each with a control to edit or forget it"
/>

Every fact it holds is on one page. Read it, correct it, delete it. In conversation it just shows up as Nova
already knowing.

<Shot
  name="chat-recall"
  width={3000}
  height={2400}
  alt="A chat where Nova drafts a renewal reply in euros and copies Marta, then explains that both rules came from something the user said in July"
  caption="Drafted in euros, Marta copied in, because you said so six weeks ago. Ask why and it names the conversation."
/>

## Workbench: numbers, not impressions

"How many times did that supplier come up, by month?" Every other assistant guesses at this and sounds
confident doing it. Nova writes a small program, runs it over its own conversation history, and gives you the
count.

The program runs on a sealed bench: no network, nothing it can write to, and a copy of the history rather than
the history itself.

<Shot
  name="chat-workbench"
  width={3000}
  height={2800}
  alt="A chat where Nova counts 31 mentions of a supplier by month in a table, and reports three conflicting lead times with the dates they were said"
  caption="Thirty-one mentions, broken out by month, plus the three different lead times that supplier has quoted you. The step named Calculated over its own history is the program it wrote to find out."
/>

## Autopilot: it keeps working after you close the tab

Find three suppliers who ship to Portugal. Watch this listing and tell me when the price moves. Draft the
monthly report the moment the numbers land.

Hand over the job, set a ceiling, walk away.

<Shot
  name="autopilot"
  width={3000}
  height={2000}
  alt="The Tasks panel: one task working with its progress note, three turns of six used and 18 cents of a 50 cent budget spent; below it a task stopped with an out of budget badge"
/>

Every row tells you where it got to, how many turns it burned, and what it spent against the limit you set.

<Shot
  name="chat-autopilot"
  width={3000}
  height={3000}
  alt="A chat where Nova takes on a supplier search as a task that runs itself, reports progress at turn three, then stops at its 50 cent ceiling with two suppliers confirmed"
  caption="Handed over at the top, reporting at turn three, then stopping itself dead at the 50c ceiling with two of three suppliers confirmed. It says exactly why the third is missing."
/>

## Splitbrain: four questions in the time of one

A due-diligence question with four independent parts used to mean four conversations. Nova sends out four copies
of itself instead, one per part, each with its own slice of the budget. They run in parallel, report back, and
Nova writes the single answer.

<Shot
  name="chat-splitbrain"
  width={3000}
  height={3400}
  alt="A chat where Nova answers four due-diligence questions at once about lease norms, flood risk, competitor rent and a port strike, with each line of enquiry visible as its own step"
  caption="Four questions, four copies of Nova, one answer. Each line of enquiry shows its own steps and its own cost."
/>

## Rewind: take the other path, keep both

Jump back to any earlier turn and continue differently. The version you left keeps its own memory, sitting
there intact, ready whenever you want it back. Two takes on the same conversation, both real, both yours.

## Correct it twice and it has learned

Third time, it applies the correction without being asked. Every lesson is written down with the conversation
that taught it, and any lesson can be rolled back.

<Shot
  name="learned"
  width={3000}
  height={2000}
  alt="The Learned panel: lessons Nova drew from being corrected, each with the skill it changed and a control to roll it back"
/>

## Routine: it shows up on time

A prompt and a rhythm is the whole setup. The FX digest at seven every weekday. The standup summary on Mondays.
The price check every hour.

<Shot
  name="chat-routine"
  width={3000}
  height={2600}
  alt="A chat where the user asks for a weekday FX briefing at 07:00 and Nova sets it up, followed by the 07:00 run flagging a 1.4 percent move"
  caption="Asked for once. The second message is the 07:00 run arriving on its own, flagging the one move over 1% and what it does to a payable due on the 29th."
/>

<Shot
  name="routine"
  width={3000}
  height={2000}
  alt="The Schedules panel listing recurring prompts with their rhythm, when each last ran and when it runs next"
/>

## Toolbelt: 14,971 tools, no signup

Web search, page reading, market data, geocoding, transcription. Flip Toolbelt on and Nova can reach 14,971
catalogued endpoints, priced per call. No API keys to hunt down, no accounts to open, no subscriptions to
cancel later. Nova finds the tool, pays for the one call, moves on.

<Shot
  name="chat-toolbelt"
  width={3000}
  height={2600}
  alt="A chat where Nova transcribes a voice note, finds two commitments, checks the address against the delivery zone and reports the 2.2 cent cost"
  caption="Nobody configured this. Nova searched the catalogue, picked a transcription tool and a geocoder, used both, and told you the whole thing cost 2.2c."
/>

<Shot
  name="toolbelt"
  width={2250}
  height={2400}
  alt="The Tools card: billed spend for the month with 82 percent of the budget used, the tools Nova may call, and the spending limits and permissions below it"
  caption="And the card that rules it: most it may spend per call, the budget window, the permissions it holds."
/>

Want Nova doing the work with its own hands? **System Shell** gives it a real shell and a real filesystem in its
own container, so it installs its own skills instead of only routing yours.

## Nine powers, and it reaches for the right one

Here is the trick nobody talks about: abilities are easy, judgement is hard. Between Recall, Pinboard, Routine,
Autopilot, Splitbrain, Workbench, Toolbelt and System Shell there are eight ways to be helpful, and exactly one
of them is right for "save this for later". Pick wrong and the agent looks like it ignored you.

So every new ability shipped with tests built to trip it. "Save this for later" must become a task, not a
schedule, not a memory. "What did I say about the Lisbon office" must search the conversation, not run code over
it. "Check the weather" must reach for a tool, not a shell.

Twenty-two traps in the suite. Nova walks past **93%** of them first try, and any release that scores worse
than the one before it does not ship.

<Shot
  name="create"
  width={3000}
  height={2400}
  alt="The create-a-Nova form with the abilities section open, all nine switched on, each with a plain description and a learn more link"
/>

Pick the powers you want when you create it. A Nova that only routes questions to your agents is a great Nova.
Turn the rest on the day you need them.

## Your money, your ceiling, no surprises

Nova buys things: searches, page reads, market data, hundreds of paid tools. So the limits live in the payment
path, not in Nova's instructions. No prompt talks past them. No clever wording gets around them.

You set the most it may spend per call, a budget per day or week or month, and the amount above which it has to
stop and ask you. The lowest limit wins, your balance sits behind all of them, and nothing is ever quietly
charged to a card.

<Shot
  name="spend"
  width={3000}
  height={2000}
  alt="The Spend panel: every paid call Nova made, with the tool, the cost and how long ago, and the conversations and tasks they belong to"
/>

Afterwards you get a list, not a lump: which tool, in which conversation, for which task, at what price. So
"why was yesterday expensive" has an answer.

## One Friday request, everything at once

This is Nova with all nine powers on. One message on a Friday afternoon: it recalled the constraints you gave
it weeks ago, asked two of your agents, counted twelve months of invoices on the Workbench, sent copies of
itself out to check two unrelated risks, put a chase on the Pinboard, set a weekday Routine, and compacted the
older turns of the conversation on the way past.

Twelve steps. One cent of paid calls. Every step named.

<Shot
  name="chat-everything"
  width={3000}
  height={4000}
  alt="A chat where Nova answers one Friday request with twelve steps: memory, two member agents, a computation over its own history, parallel research with paid calls, a task, a schedule and a context compaction, then a follow-up turn"
  caption="Twelve steps for the big ask. Directly underneath, the same Nova routes a small question straight to one agent. Power on tap, not power on display."
/>

## Run one per job

<Shot
  name="fleet"
  width={2400}
  height={3000}
  alt="The Nova index: several Novas, each card showing its agents, its abilities as icons, spend for the period and a daily spend chart"
/>

Nova is not one assistant everybody shares. Spin up one per job, each with its own agents, its own powers and
its own budget, and the bill splits the same way.

## New in 0.52.0: counting what the text never names

Ask how many dice rolls happen in a game transcript. The real answer is 84, in a document where the word "roll"
appears 40 times, because people write rolls as "Matt: 18." Search finds nothing. A frontier model handed the
whole file answered 39, roughly the share of it that got read.

Nova now works that file the way you would with a spreadsheet open: split it into slices, read every slice,
write down each hit, add the list up in code. The arithmetic never happens in its head. If a slice would not
read, it says so and calls the total a floor.

Same 30 questions, same answer key: **0.167 before, 0.503 after.** A 3x jump, 24 of 30 questions improved, none
worse. And it stays cheap when cheap works: on the biggest file in the set, 4.3 megabytes, Nova nailed the exact
answer with a two-line program and never touched the expensive path.

## Six releases of receipts

Every ability on this page was scored before it shipped, and the scores are committed to the repo with the
prompts and the answers Nova gave. Nothing goes out on vibes.

<BenchLine
  caption="Six releases of new powers, and nothing that worked before stopped working. Solid lines are public benchmarks anyone can run against us. Dashed lines are our own tests, which exist because no public benchmark can see two of Nova's abilities colliding."
  source="bench/results/*.json: one committed file per release, every case with its prompt and Nova's answer, three runs per case at temperature zero"
  series={[
    {
      name: "asks the right agent",
      comparable: false,
      points: [
        { x: "0.46", y: 0.889 },
        { x: "0.47", y: 1.0 },
        { x: "0.48", y: 1.0 },
        { x: "0.49", y: 1.0 },
        { x: "0.50", y: 1.0 },
        { x: "0.51", y: 1.0 }
      ]
    },
    {
      name: "picks the right ability",
      comparable: false,
      points: [
        { x: "0.46", y: 0.933 },
        { x: "0.47", y: 0.885 },
        { x: "0.48", y: 0.967 },
        { x: "0.49", y: 0.853 },
        { x: "0.50", y: 0.93 },
        { x: "0.51", y: 0.932 }
      ]
    },
    {
      name: "finds a tool for the job",
      comparable: false,
      points: [
        { x: "0.46", y: 0.867 },
        { x: "0.47", y: 0.833 },
        { x: "0.48", y: 0.833 },
        { x: "0.49", y: 0.8 },
        { x: "0.50", y: 0.8 },
        { x: "0.51", y: 0.867 }
      ]
    },
    {
      name: "recalls across sessions",
      points: [
        { x: "0.46", y: 0.599 },
        { x: "0.47", y: 0.636 },
        { x: "0.48", y: 0.604 },
        { x: "0.49", y: 0.601 },
        { x: "0.50", y: 0.626 }
      ]
    },
    {
      name: "counts over long transcripts",
      points: [
        { x: "0.47", y: 0.211 },
        { x: "0.48", y: 0.144 },
        { x: "0.49", y: 0.178 },
        { x: "0.50", y: 0.178 },
        { x: "0.51", y: 0.167 },
        { x: "0.52", y: 0.503 }
      ]
    }
  ]}
/>

Every release added something Nova could not do before and had to clear the bar the last one set. That is the
only promise worth making about an agent that keeps growing: the new thing works, and the old things still do.

<CTA href="/nova" label="Build a Nova">
Pick the agents you already have, give it a name, press deploy. All nine powers are switches. Flip them the day
you want them.
</CTA>