Release Notes

This release makes Target Intelligence’s clinical progression predictions sharper and dramatically easier to trust at a glance.

New: Indication-Specific Predictions

Clinical progression scores are no longer a single number for a gene — they’re now specific to the exact disease you’re evaluating. Ask about a target for one indication and get a prediction built for that pairing, not a generic average across everything the gene has ever been studied for.

New: Percentile Rank Replaces Raw Probability

Instead of a raw probability that could look alarmingly low even for a target with an approved drug already on the market, every prediction now leads with a percentile rank — “this pairing ranks in the top 10% of everything the model has seen” — a number you can trust and compare at a glance, phase by phase.

Improved: Clinical Landscape Chart

The Clinical Landscape chart in Target Explorer has been redesigned around this new percentile rank. Pick exactly the indication you care about from a ranked list, and see its own percentile for Phase 1, 2, and 3 trials side by side — no more hunting for the same number repeated across a long, scrolling chart.

Improved: Reports Reflect What You Actually Asked

When you name a specific disease alongside a target, Target Intelligence reports and their clinical-progression charts now reflect that exact disease — not just whichever indication the platform would have guessed on its own.

And as always

We’ve continued hardening the platform’s prediction infrastructure so the numbers you see keep getting more reliable as the product gets more capable.

This release focuses on smarter research recommendations, deeper target-discovery scoring, a new bispecific antibody design capability, and a faster, more accurate reporting pipeline across every research agent.

New: Cross-Agent Recommendations

Every report can now suggest the right follow-up analysis automatically. After a Deep Research or Competitive Landscape report finishes, the platform reviews what was discussed and offers one-click hand-offs to Target Intelligence, Drug Repurposing, Competitive Landscape, Biomarker Discovery, or Hypothesis Generation — so you go from “interesting finding” to “next analysis” without re-typing context.

New: Smarter Target Discovery — Novelty vs. Validation

Target Intelligence can now rank candidate genes by how novel they are, not just how well-validated. A new Novelty vs. Validation chart plots every candidate on two axes — how under-explored a target is against how strong the causal evidence is — so you can find genuinely first-in-class opportunities instead of re-discovering the same well-known targets everyone else is already chasing. Candidate genes are now pulled from multiple independent sources (OpenTargets, DisGeNET, knowledge-graph mining, and literature/GWAS text-mining) and merged by consensus, broadening discovery beyond any single database’s blind spots.

New: AI-Powered Clinical Progression Prediction

Target Intelligence reports now include a machine-learning-based prediction of clinical progression likelihood for each candidate, alongside a single combined GO / CONDITIONAL-GO / NO-GO recommendation — a clear, defensible verdict instead of a wall of disconnected scores.

New: Bispecific Antibody Design

The Antibody Design agent (AADA) now supports bispecific antibody design across all 17 industry-standard formats, in addition to standard monoclonal design — expanding computational antibody discovery to dual-targeting therapeutics.

New: Interactive Visualizations in Research Reports

Deep Research, Competitive Landscape, and Biomarker Discovery reports now render interactive charts directly inline — knowledge-graph relationship diagrams, clinical trial landscape breakdowns, and predicted-vs-actual outcome timelines — so you can explore the underlying evidence visually instead of reading raw tables.

New: Floating Platform Assistant

A draggable, always-available chat assistant now floats over every page in the BioMedAI app. Ask it questions about any agent, any report, or the platform itself without losing your place in what you were doing.

Improved: Report Accuracy & Citations

We tightened citation handling across every research agent — Deep Research, Competitive Landscape, Biomarker Discovery, Drug Repurposing, and Pharmacovigilance now produce consistent, correctly formatted citations (author, journal, year) for both literature and clinical-trial sources. Clinical trial counts shown in charts now always match the trials listed in the accompanying table.

Improved: Faster, Streaming Reports

Biomarker Discovery, Competitive Landscape, Drug Repurposing, and Pharmacovigilance reports now stream their analysis live and run research steps in parallel, so you see meaningful progress section-by-section instead of waiting for the entire report to finish.

Improved: Hypothesis Generation for Any Disease

The Hypothesis Generation agent is no longer limited to a single disease area — it now works disease-agnostically across any biomedical topic, with deeper knowledge-graph-backed evidence.

Improved: Smoother Sessions & Navigation

Your login session now refreshes silently in the background, so you’re far less likely to be interrupted mid-task. The AI Tools sidebar has been reorganized for clarity, with new entries for Omics Intelligence and KO Intelligence, and faster gene-search autocomplete throughout.

And as always

We’ve continued hardening the platform’s infrastructure — deployment, security, and reliability improvements across every agent — so the product you depend on keeps getting steadier as it gets more capable.

This release introduces in-app feedback, so you can tell us what's working (and what isn't) without leaving a report.

New: In-App Feedback

Rate and comment on any report or workflow step directly from the chat view — a quick star rating plus optional comments, reviewed by our team through a new admin feedback dashboard. We also added WeDaita's GraphRAG MCP server as a default knowledge source for richer evidence retrieval.

A focused round of polish to the chat experience and report views, plus deeper streaming support in Competitive Landscape.

New: Cleaner Activity Views

Internal tool-call/technical details are now hidden by default in production across Deep Research, Target Prioritization, Drug Repurposing, Biomarker Discovery, and Competitive Landscape activity panels — reports stay focused on findings instead of internal research chatter.

Improved: Chat Experience

Interactive prompts, clearer progress display, and suggested follow-up questions make it easier to steer a conversation, alongside fixes to how Hypothesis Generation's activity steps render.

Improved: Competitive Landscape Streaming

Competitive Landscape reports now stream structured analysis live after each research tool call, with deeper knowledge-graph integration behind the scenes.

The first coordinated release of the BioMedAI platform — eleven purpose-built research and design agents, a unified chat experience, and the credit and deployment infrastructure to run them at scale.

New: Antibody Design (AADA)

Autonomous antibody design across four modes — binder, CDR redesign, VH+VL, and macrocycle — orchestrating RFdiffusion, CDR-H3 identification, and AlphaFold2-based structural QC (ipTM/PAE confidence gating). Failed candidates automatically trigger a QC-driven redesign loop, and you can restart or resume a workflow from chat at any phase while GPU jobs run asynchronously with live status updates.

New: Deep Research

Multi-agent biomedical research reports with mandatory inline citations and structured bibliographic metadata, plus podcast and presentation export for sharing findings.

New: Competitive Landscape

Competitive analysis reports across drugs, targets, and trials with inline citations and an at-a-glance spider chart summarizing the landscape.

New: Biomarker Discovery

Biomarker identification reports synthesizing genomic and clinical evidence, visualized with a spider chart for quick comparison across candidates.

New: Drug Repurposing

A supervisor-coordinated team of five parallel specialist agents evaluates repurposing candidates using network-proximity scoring against a protein-protein interactome, with a spider chart summarizing each candidate's composite score.

New: Target Prioritization

Target intelligence reports built on real API data across every table, with PubMed-backed citations and a streamlined report header and methodology section.

New: Pharmacovigilance

Automated drug safety signal analysis, integrated with the platform's credit and reporting infrastructure from day one.

New: Hypothesis Generation (WeDaita AI Scientists)

An autonomous hypothesis-generation agent, integrating PanKgraph knowledge graphs and MindsDB-backed AIRR TCR/BCR repertoire data, with a glassmorphism chat interface and persistent thread history.

New: Unified Chat Platform & Landing Experience

One agent-aware chat hub routes every conversation to the right specialist, with model selection, file upload, and live GPU job activity timelines. The new landing page introduces the platform's three-stage pipeline — DBRA (target research) → ADDA (drug design) → CTIA (clinical trials) — alongside membership tiers and a referral program.

New: Platform Infrastructure

A unified API gateway routes requests across every backend agent, backed by a shared credit-accounting and authentication library so usage is tracked consistently no matter which agent you're using.