Top AI Agent Projects : Looper, Hyperresearch, Agent Inbox, Cedric & Click Transcript, AI Summary & Key Points

Agent Inbox is an email, Slack, and Telegram inbox for AI agents, built and operated by Novu. Its CLI creates an inbox and email address for an agent, lets it send and reply to conversations, and keeps messages from the three channels in one thread. Agents can use structured prompts such as ask, approve, or choose to request a response, with options for asynchronous requests, timeouts, expiration periods, JSON output, contact management, and MCP access. The service is designed to work with coding and browser agents, personal assistants, product agents, and support workflows; its site states that no signup or credit card is required for setup and that Novu's infrastructure has SOC 2 Type II and ISO 27001 certifications.

Cedric AI is an AI workplace and meeting assistant from a Swiss-based company in Zurich for Google Meet, Zoom, and Microsoft Teams. It can join as a listener, moderator, or stand-in; present decks or workspace pages; answer live questions using connected tools and the web; keep time; take notes; record searchable transcripts, decisions, summaries, and action items; and attend meetings on a user's behalf based on a brief and behavior profile. Before a call, users can provide goals, topics to watch, questions, and boundaries. Cedric drafts follow-up work such as Gmail or Outlook emails, CRM updates, Notion documents, tasks, reports, and campaign content for review and approval, and connects to tools and channels including Slack, Discord, Telegram, email, calendars, HubSpot, Salesforce, CRM systems, GitHub, Notion, Sheets, and Drive. It uses shared workspace memory to answer questions, book meetings, update systems, and coordinate scheduled or event-triggered work. Outward actions show a preview for approval and are recorded in an activity log; the site says customer data is not sold or used to train models and that the service is GDPR compliant.

ClawsGO Science is an AI research agent for scientific work. It takes a research task in natural language, searches academic sources, reads papers through a cloud browser, analyzes data, writes and runs code, and produces cited reports, datasets, figures, or LaTeX manuscripts on cloud compute. Its runs can continue unattended for hours and expose the search history, commands, file edits, and intermediate steps for review and replay. The service also supports switching between Claude and OpenAI models and, according to the video description, connecting a lab GPU over SSH.

Click is a managed agents API for running serverless AI agents in isolated sandboxes. An application submits work through an API call, while Click manages execution, durable session state, tool access, and streamed events or webhook notifications when background work completes. Agents can use browser, web search, an OS/code sandbox, and application-provided tools through MCP servers or HTTP endpoints; sessions can specify tool permissions and credentials. The service supports any model with customer-provided API keys and handles file uploads and downloadable artifacts.

Display is an AI documentation workspace for publishing and collaborating on HTML and Markdown files. Agents such as Claude Code, Claude, ChatGPT, Codex and Cursor connect through its MCP server, CLI or API and publish documents to private, shareable URLs; people can comment on lines, edit pages and share them through groups and permissions. Each save or publication creates a new version at the same URL, with earlier versions available for restoration on eligible plans. Spaces, folders and search organize reports, dashboards, specifications, prototypes, slides, documents and related files, including images, PDFs and data. Google, Microsoft or one-time-password sign-in controls access. Display also offers private-beta features for building internal applications and connecting data sources, with free, Solo, Team, Business and Enterprise plans that charge for editors or contributors rather than viewers.

Embage is a no-code platform for building AI chat and browser voice agents for customer support, sales, and feedback. Agents can run on a website or Telegram, answer from uploaded documents and web pages, save leads, tickets, and feedback in a datastore, and hand conversations to human team members. Its agents use plain-English sub-agent workflows to call connected tools during conversations, with integrations including Shopify, HubSpot, Salesforce, Google Calendar, Google Meet, Gmail, Zendesk, Slack, and Stripe. After a conversation, Embage scores it from 0 to 100, summarizes it, identifies action items and knowledge gaps, and records the result. The platform supports more than 360 connected apps, offers a free plan, and is available in early access.

ernd is a business credit card for U.S. companies that spend on AI services. Card purchases at AI vendors, including OpenAI and Anthropic, earn 4% back, while other purchases earn 1.5%; certain negotiated provider offers may provide additional rewards or benefits. ernd negotiates partnerships with AI providers and passes benefits such as usage limits, provider discounts, and priority GPU or inference capacity to cardholders. Points can be redeemed for cash back or used for travel and participating providers. Card requests are free with no annual fee stated, but final terms apply and requesting a card does not guarantee approval.

Framer is a web-based design and website-building platform that combines visual design, prototyping, and site creation with built-in hosting, CMS, analytics, security, and SEO. Its AI design agent generates page designs and websites from textual prompts on a visual canvas; the generated designs remain editable and can be deployed through Framer’s hosted SaaS.

Hera is an AI-native creative canvas for generating images, designs, and motion graphics on one canvas. Users can brief multiple AI agents at the same time, generate and compare results, and refine outputs as editable layers rather than flattened images; the canvas supports design work, animation, and launch-film creation. The video describes Hera as offering an API and MCP, while the product page says it provides editable-layer generation and tools for directing, iterating on, and animating designs.

Hyperresearch is an MIT-licensed Python command-line research harness that extends Claude Code with an agent-driven, persistent research knowledge base. It classifies a query into light, full, or opt-in dissertation work and runs a tier-adaptive pipeline covering query decomposition, source search and fetching, contradiction and locus analysis, depth investigation, evidence digestion, multi-angle drafting, synthesis, adversarial criticism, targeted gap fetching, surgical patching, citation checks, and readability review. The pipeline stores fetched sources as Markdown notes with a rebuildable SQLite index, provenance links, lifecycle statuses, full-text or optional semantic search, source-quality and retraction metadata, and resumable per-run manifests. It can fetch PDFs, seek disclosed open-access replacements for thin or blocked academic sources through Unpaywall and Europe PMC, and apply structural checks for citation bindings, quote integrity, provenance, retractions, and patch-only editing. The vault can also be accessed through an MCP server or a dependency-free local web UI. It is installed with pip for Python 3.11–3.13 and is intended to run with Claude Code; model assignments and research scale can be configured per project.

Ideogram Precise Edit is an AI image-editing tool that changes a selected area of an image without regenerating the rest. It is available in the Ideogram app and through an API, and the video describes edits at resolutions up to 8K.

Looper is an open-source AI teammate platform in which each looper owns one business metric, such as organic clicks, response time, post reach, qualified leads, or unreconciled transactions. A looper connects to business tools such as Gmail, HubSpot, Stripe, Search Console, GitHub, Linear, Buffer, and other services through Composio; it plans work on a schedule, runs skills, updates a Kanban board, moves the metric, and conducts a weekly retrospective. Researching, drafting, and planning can happen independently, while sending, publishing, paying, and deleting require approval. Looper can run using a user's ChatGPT or Grok subscription or API key, with paid plans adding Claude credits, and the page also describes self-hosting through Vercel.

ManuAGI & AutoGPT Tutorials is an email newsletter covering AI, AGI, and AI-agent project updates, with new updates promoted weekly. It is available through a free subscription page published through beehiiv.

Modelpeers is a tracker for AI agents, models, and companies. It records sourced releases, price and limit changes, retirements, breaking changes, incidents, security issues, quality updates, and benchmarks, and provides pricing, usage-limit, and head-to-head comparisons. Users can follow selected entities to receive weekly alerts; the service also offers an RSS feed.

Precise Edit is Ideogram's AI image-editing tool for localized changes. It uses a context window around a marked area to edit images at their original pixels while leaving the rest of the image untouched, avoiding seams and color mismatches; the context window can cover an edited area up to 2K within images up to 8K. It supports recoloring products, repairing logos and fine detail, replacing or removing objects, and using reference images, and is available in the Ideogram app and through an API.

Riverside Agents is a team of AI specialists built for video production inside Riverside. The Producer prepares questions, scripts, studios and recording sessions; the Editor turns raw recordings into finished episodes by suggesting cuts, tightening pacing, cleaning audio, switching layouts and adding branding and captions; the Clipper finds moments for social posts; and the Marketer writes captions, fills a publishing calendar and schedules posts. The agents work natively with Riverside’s recording and production tools, and can repurpose existing recordings or support webinars, testimonials and branded business video. Users can direct the workflow in plain language, review or change the agents’ work on web or mobile, and approve everything before publication. Riverside says the agents are available on every Riverside plan, with capabilities depending on the plan and some AI features using credits; a 14-day free trial is offered.

SKI is a voice-first coding harness for macOS from Pattern AI Labs. It lets users speak to coding agents and hear their responses, while running agents such as Claude Code, Codex, Cursor, OpenCode, Antigravity and pi side by side. Each chat receives its own Git branch and worktree, with live diffs, cross-project status, agent handoffs and voice-triggered pull requests. SKI reconstructs spoken intent using project context, shows the interpretation before sending it to an agent, and requires confirmation for irreversible actions. It can chain agents, such as having one agent plan a task and another implement the plan, and supports routing work among users' existing model and agent accounts. Project code, chats and agent execution remain on the Mac; voice processing uses the cloud and the site says the voice is not stored. The full harness is available for macOS with free and paid plans, while Windows and Linux are listed as forthcoming.

Specsight is a product-observability platform for keeping a company informed about software changes. With read-only access to a codebase, it builds a product changelog, a map of connected user journeys and features, and a product specification describing behaviors from the code; these artifacts are updated as the product changes. Users can ask questions about the product in Specsight or through MCP, and can generate release notes and weekly product updates from detected changes.

telemetry.dev is an LLM observability platform for AI agents built on OpenTelemetry. It traces model calls, tool steps, retrievals, and MCP calls in a shared trace, exposing token usage, cost, latency, errors, and per-step timing. Applications can connect through TypeScript or Python SDKs or an existing OTLP/HTTP exporter. It provides filtering and search by status, environment, framework, or session, plus email alerts for costs, budgets, agent loops, and model or tool errors. Prompt capture can be disabled for individual environments.

Widgeteer is an AI widget builder that creates working forms, quizzes, booking tools, calculators and other embeddable widgets from plain-English descriptions. Users can preview and submit the real widget, edit questions, answer types, messages, colors and layouts by hand, or request changes in plain English. Widgets can be embedded on any website as code, or added through Shopify and WordPress integrations. Submissions can trigger email notifications or be routed to Zapier, n8n, Make, webhooks, Airtable, Telegram and other services.

Winneradspy is an ad creative intelligence platform for mobile apps and games. It lets marketers search an app or developer and browse the history of its install, video, banner, and other ad creatives, filtered by platform, date, format, and win rate. The platform monitors competitor ad activity and alerts users when creatives or campaigns are launched, paused, or changed. Winneradspy classifies creatives as Winner, Loser, or Learning and assigns each a 0–100 Winner Score. The score breaks down the creative's hook, visuals, emotions, messaging, and call to action; its classification uses signals such as ad runtime, platform reach, creative reuse, and engagement indicators. The page identifies Facebook, Instagram, and the wider Meta ad network as its tracked advertising platforms, and also says that data is pulled from Meta Ads, Google Ads, the App Store, and Google Play.

Xmind is an online and desktop mind-mapping and project-planning application for organizing ideas into visual structures. Xmind Agent turns a sentence, file, link, or conversation into a structured mind map, then lets users debate, refine, expand, reshape, and develop it through conversation, identify missing areas, and break goals into prioritized tasks with working-day planning. Maps can be presented or organized as timelines, matrices, Zones, and Gantt-chart project plans, with options to export, print, or collaborate. Xmind lists integrations with Claude, Claude Code, ChatGPT/Codex, Cursor, and Notion, and offers individual, business, enterprise, and education plans.