No-Code Agentist

Nous Research just open-sourced what Jane Street pays quants $500K+ to build. It's called Hermes Agent. Deploys in 30 minutes on a $5 VPS. One trader turned $300 into $123K trading weather markets with it. Every trade on-chain. No code required. https://t.co/KqFHmNJM7p

the bankr agent now natively supports skills. just uploaded two new skills you can tell bankr to install for you. > token scam analysis: https://t.co/yKrafYIVO7 > twitter agent (set up an agent to post on twitter powered by bankr): https://t.co/0lvcDPOhLH terminal: https://t.co/eDe9nH3pWJ

the marketing knowledge layer for agencies the single highest leverage thing you can do for your agency agents is centralize your data. structure it, make it searchable, and put it in one knowledge layer that every agent reads before it starts working. the better the context you give them, the better the output you get back. the performance data is where most of the value sits: > campaign results across every client, channel, format, and vertical with connected data showing which headlines converted, which CTAs drove clicks, which creatives burned budget > ad spend and ROAS history per client, per channel, per quarter, so when you pitch a new prospect in the same vertical you pull the real numbers from the last three projects > SEO rankings and traffic patterns showing which blog topics drove signups, which pages ranked in 30 days vs which took 6 months > email and outbound metrics broken down by subject line pattern, reply rate by persona type, conversion by sequence length, all connected across your entire client base then theres the operational data that most agencies lose to slack threads and email chains: > client call transcripts and summaries auto-captured from every meeting, every objection, scope change, and "what we tried last time" is searchable > proposal history with outcomes (won, lost, ghosted), the actual language that closed deals, and heatmap data showing which sections clients spend the most time reading > per-client brand foundations holding voice rules, audience, positioning, visual style, words they never use > QA logs tracking which delivery issues repeat and how they were solved every employee also gets a private channel where they can dump whatever is on their mind. ideas, observations, things they noticed on a client call, a campaign approach they want to try, feedback they didnt want to push into the main chat because it would get buried. all of it flows into a separate folder in the knowledge layer. an agent can surface patterns across everyones notes, pull up ideas that would have been lost in slack, and flag insights about the business that nobody connected on their own. it becomes a second brain for the entire company. none of this needs a documentation project. call recordings summarize themselves, ad platforms export data, CRM logs interactions automatically, employee notes go in as raw thoughts. the data flows in from work your team is already doing. once the knowledge layer has depth, agents start producing work that would take a human hours of context gathering. a media buyer agent pulls performance data from the last 5 clients in the same vertical before building a new campaign, already knowing which audiences and bid strategies worked. a proposal agent writes SOWs using winning language from past proposals in the same industry, references actual results like "we drove 340% ROAS for a similar DTC brand in Q3" with the data to back it, structured around the sections that get the most attention in heatmap data. onboarding works the same way, the agent matches a new client against similar past projects so a new account manager is productive day one. reporting agents build client decks from their specific KPIs and historical context, flag performance drops, draft the summary before the AM even opens the file. content agents write in each clients brand voice. research agents monitor competitors, scrape creative from Meta Ad Library, feed it back into the knowledge layer so the next brief starts sharper. every project adds to the knowledge layer, and while every client is different, the patterns help. especially when similar clients come through later, your agents already have performance data from comparable campaigns instead of starting from scratch. agencies that have been good at documenting, keeping clean data sets, and organizing results in a structured way are about to have a massive advantage. their agents get better context from day one, which means better output from day one. the chaotic ad-hoc agencies running on slack threads and scattered google drives will struggle to get the same quality out of the same models.

🤖 Not all AI models are built equal for Hermes Agent. After 1 month of testing MIMO V2 Pro, GPT 5.4, Kimi 2.5, DeepSeek 3.2 & more. We ranked every model by Orchestrator, Executor & Auxiliary roles. Save this before you pick your next model. 👇 https://t.co/fFbNztunNm