
Pushing the frontier @Mercor Prev. co-founder @conciergedotai 𦾠|| VC @Revolution š || Product @Stryker_NV š§ || Research @ J&J š«
San Francisco, CA
Joined February 2016
- Tweets347
- Following2.5K
- Followers593
- Likes919
Pinned Tweet
Some BIG news to share⦠I left venture capital to do the ā0 to 1ā thing. It was time to get in the arena. And Iām so proud to share that today, we got to 1. @askconciergeai is LIVE on Product Hunt š
@rohanagrawal @cssetian
producthunt.com/posts/concieā¦
Ayush Jain retweeted
2012: acquire company for users
2021: acquire company for engineers
2026: acquire company for training data
Congratulations to the @Harvey team on releasing their Tenet model.
Human expert data is crucial for post-training work. Weāre proud to have collaborated with @calvincongelado @vtrengarajan @ItsJulioPereyra @nikogrupen @gabepereyra, the Harvey research team, and hundreds of Mercor experts on this project.
Introducing Tenet, our first model post-trained for legal.
Tenet is a Kimi K3 base that we post-trained with @FireworksAI_HQ on a corpus of publicly available legal data, synthetic data, and human expert data simulating long-horizon legal work.
Training increases Tenet's all-pass rate by 82% on LAB and 22% on LAB Contracts relative to the Kimi K3 base model. It achieves state-of-the-art performance on LAB Contracts and places second on LAB.
These gains generalize to other leading agentic benchmarks including @mercor's Apex Agents - Corporate Law, @crosbylegal's Redline Bench, and @scale_AI's Professional Reasoning Bench.
Tenet is also optimized for token efficiency, operating at less than a fourth the cost of leading foundation models.
We additionally post-trained three specialist models for Tenet to use as subagents:
1) M&A Diligence: post-trained with @baseten on our LAB Diligence environment in an RLM harness, this model is optimized for high-scale, long-horizon tasks.
2) Review Tables: trained with @appliedcompute on our Review Table environment, this model is state-of-the-art and cost-effective at high-volume document review and structured data extraction.
3) Firm Knowledge: trained with @EngramLab on our synthetic law firm environment, this model is optimized to learn and search over a firm's knowledge via memory and structured notes.
More details on model training, environment design, benchmarking, results, and more in the article by @gabepereyra below.
What's next for Harveyās research?
- Scaling LAB to more jurisdictions, practice areas and workflows
- Scaling compute to bring new generalist models and capabilities to Harvey
More to come soon.
Ayush Jain retweeted
+1 for "context engineering" over "prompt engineering".
People associate prompts with short task descriptions you'd give an LLM in your day-to-day use. When in every industrial-strength LLM app, context engineering is the delicate art and science of filling the context window with just the right information for the next step. Science because doing this right involves task descriptions and explanations, few shot examples, RAG, related (possibly multimodal) data, tools, state and history, compacting... Too little or of the wrong form and the LLM doesn't have the right context for optimal performance. Too much or too irrelevant and the LLM costs might go up and performance might come down. Doing this well is highly non-trivial. And art because of the guiding intuition around LLM psychology of people spirits.
On top of context engineering itself, an LLM app has to:
- break up problems just right into control flows
- pack the context windows just right
- dispatch calls to LLMs of the right kind and capability
- handle generation-verification UIUX flows
- a lot more - guardrails, security, evals, parallelism, prefetching, ...
So context engineering is just one small piece of an emerging thick layer of non-trivial software that coordinates individual LLM calls (and a lot more) into full LLM apps. The term "ChatGPT wrapper" is tired and really, really wrong.
Ayush Jain retweeted
Afternoon deep dives included:
ā© All steak, no sizzle ways founders are leveraging AI with w/@ayush_concierge & @ijablokov
š Creating cultures and teams built for change w/ Ted Souder & @andycloyd
š¬ Breakouts for founder-to-founder pressure-testing and candor
Setting this up isnāt as easy at it should be. We solved that with askconcierge.ai - in <30 seconds you can connect your apps in 2 clicks, pick your favorite model (o3, Claude, deepseek, Gemini etc), and ask anything.
Very well stated š AI at scale is a force multiplier for every person, business, and market.
The thing some people get wrong in thinking about AI is imagining the world as a zero sum system where we can only create a certain amount, and so AI creating means taking something away from people working.
This is entirely the wrong way to think about AI, and the economy generally. We should think about AI as a technology that democratizes access to some of the hardest and most complex things in the world so we can do way more. What happens as a result?
Think of all the filmmakers or writers without access to big budgets that want to produce new ideas. The scientists without extensive labs that want to do research. The marketers without 10 agencies supporting them that want to scale to more markets. The engineers without an entire team to build the next set of features or clean up their bugs.
The vast majority of potentially great work in the world is currently limited by the access to the capital or human resources that make all these things possible. And even when they do have access to those things, everything is just *slow*.
A breakthrough new drug costs billions and can be a decade of work. A blockbuster movie costs hundreds of millions and takes years to make. A marketing campaign takes millions and months to execute. And so on.
With AI, we start to unlock an ability to accelerate all of this work, and make it accessible to far more people, in a way that wasnāt comprehensible before.
And no, this doesnāt mean that everyone will start to do all of these things to point where thereās no value left. In every one of these fields thereās a very long tail of complex work and judgment requirement to produce real value. This gap will always separate consumers from producers. It just means that *more* people will be able to do these things when theyāre motivated to do so.
In the process, many industries and even labor forces will expand as a result. Making a field easier doesnāt tell you anything about the ultimate curve of demand; all that matters is can that category offer more and more value for the world as it becomes more broadly available. For instance, weāve seen a ~10X increase in the number of software engineers in the US from 1980 to 2020, and software is far more valuable today than it was then by at least 100X. Because the usefulness of software has gone up exponentially in that time period, partly as a result of how much easier we made it.
The same will be true for many other AI-enabled categories where thereās almost no shortage of demand potential, like healthcare, life sciences, coding, marketing, entertainment, education, manufacturing, logistics, legal work, and more.
AI will definitely change almost every sector as we know it, but not always in the ways some imagine.
Ayush Jain retweeted
AI has changed how we search and get stuff done, and now that extends to your work day.
We were already asking questions about our apps using Conciergeās AI enterprise search engine - weāre excited to add realtime, shareable knowledge cards you can trust. Great to partner with @rohanagrawal, @ayush_concierge, and Chris Setian and the @askconciergeai team. Check it out!
Enterprise Apps? ā
Web Search? ā
Trusted Knowledge Cards? Now here!
Weāre excited to make @TakoViz's trusted knowledge search available to all Concierge users. Tako goes deeper than traditional web search by visualizing trusted vertical data directly in Concierge. On Concierge, you can now:
- Understand whether earnings surprised: āNvidia earningsā
- Put the Wiz acquisition in context: āGoogle acquisitionsā
- Review tomorrowās matchup: ānext Warriors gameā
- Check the weather: āSan Francisco weatherā
- Get context about the news: āhow much does Ukraine spend on its militaryā
And so much more. Check it out and let us know what you think!
Ayush Jain retweeted
Enterprise Apps? ā
Web Search? ā
Trusted Knowledge Cards? Now here!
Weāre excited to make @TakoViz's trusted knowledge search available to all Concierge users. Tako goes deeper than traditional web search by visualizing trusted vertical data directly in Concierge. On Concierge, you can now:
- Understand whether earnings surprised: āNvidia earningsā
- Put the Wiz acquisition in context: āGoogle acquisitionsā
- Review tomorrowās matchup: ānext Warriors gameā
- Check the weather: āSan Francisco weatherā
- Get context about the news: āhow much does Ukraine spend on its militaryā
And so much more. Check it out and let us know what you think!
Couldnāt agree more š @askconciergeai
Ayush Jain retweeted
ConciergeĀ is the firstĀ connected AI assistantĀ that lets you talk to your SaaS apps in real-time ā”ļø
It can read & write to your favorite tools like Gmail, Slack, Jira, Notion, Linear, Confluence, HubSpot, Attio, Salesforce, and Airtable. Ask Concierge anything and watch the magic āØ
Try it for free: askconcierge.ai
An important, thought provoking paper by @alexandr_wang @ericschmidt @hendrycks introducing a deterrence concept called āMutual Assured AI Malfunction (MAIM)ā
nationalsecurity.ai/
Ayush Jain retweeted
Revolutionize Your Workflow with Concierge AI ā”š¤
Say goodbye to tedious app switching and hello to the future of productivity. Concierge AI is the game-changing assistant designed to seamlessly integrate across your favorite software tools, allowing you to automate tasks and extract valuable insights in real-time.
Hereās how it works:
1.Create your Concierge account and start for FREE.
2.Connect your go-to tools ā Gmail, Slack, Jira, Notion, HubSpot, and more.
3.Ask Concierge for tasks like:
ā¢āSummarize my unread emailsā
ā¢āCreate a Jira ticket based on Slack feedbackā
ā¢āResearch these prospects from Salesforceā
4.Watch as your AI assistant instantly pulls data and performs actions across all your connected apps.
Concierge AI is not just a tool; itās your personal productivity partner. By streamlining cross-app workflows, it saves you time and gives you a competitive edge.
Pro tip: Start with the apps you use daily and scale up as you see the power of true AI integration. š
š Get started now: askconcierge.ai/
#ArtificialIntelligence #ProductivityHacks #AI #WorkflowAutomation #TechInnovation #BusinessTools #Efficiency #ConciergeAI #AIForBusiness #FutureOfWork @Nathcohen
Ayush Jain retweeted
Meet Coincierge (by @ayush_concierge): The ultimate productivity boost you didnāt know you needed! š
What it does šš»
Ayush Jain retweeted
Firm client @askconciergeai, the first connected #AI assistant that "talks to your favorite software tools, in real-time," announced its launch from stealth and disclosed an initial investment round! Learn more at: wsgr.com/en/insights/wilson-ā¦
Hello World! Meet Concierge.
Concierge is the first connected AI Assistant that talks to your favorite software tools, in real-time. Just 2 clicks to set up and experience the magic šŖ
Check us out on Product Hunt today š§” producthunt.com/posts/concieā¦
Ayush Jain retweeted
Huge congratulations to our very own Ayush Jain and Rohan Agrawal at Concierge AI on their Product Hunt launch today! Check it out and share your thoughts: lnkd.in/dwQVhDBg
Some BIG news to share⦠I left venture capital to do the ā0 to 1ā thing. It was time to get in the arena. And Iām so proud to share that today, we got to 1. @askconciergeai is LIVE on Product Hunt š
@rohanagrawal @cssetian
producthunt.com/posts/concieā¦
Ayush Jain retweeted
Work today = endless context switching across SaaS tools. @askconciergeai fixes that.
This AI assistant doesnāt just generate textāit acts, working across your go-to apps in real-time.
As they launch on @ProductHunt today, we're proud to back them. š
pennyjar.com/goodbye-contextā¦
Ayush Jain retweeted
Stoked to back @askconciergeai!
š§µ1/4 Thrilled to announce our investment in @askconciergeai - a unified AI interface for fragmented tools! š
cc @GradientVC






























