Use Benchling AI chat

Benchling AI has a unified chat interface that allows you to explore your data, generate documents, and conduct in-depth research from a single entry point. With each query, Benchling will handle the complexity of selecting the right tools and skills to complete a task. You use natural language prompts that specify what data you want to use as inputs and what you want as outputs and the interface routes you to the appropriate mode automatically. Benchling AI supports multi-turn chats and file attachments and includes citations and chip links where possible so you can verify sources quickly. 

Additionally MCPs allow Benchling AI to access data from external systems like Microsoft Teams, SharePoint, Slack, and other platforms that support the Model Context Protocol (MCP). This extends Benchling AI’s ability to provide comprehensive answers by incorporating information from multiple sources across your organization's data ecosystem. MCPs work by establishing secure connections between Benchling and external MCP servers, allowing Deep Research to retrieve relevant information while respecting your authentication and permissions in those systems.

This is a Benchling AI feature. See the AI at Benchling page for information on how to ensure that this and other AI-based features are provisioned to the appropriate users on your tenant.

This is in preview, so keep in mind that it is a work in progress. While in preview, this can be enabled by a Tenant admin. 

For more information on how to do that, see the Getting started with Benchling AI article. 

Access Benchling AI chats

Each time you want to create a Benchling AI chat, you will need to create a new chat where you will enter your prompt. To create a new chat, follow the steps below: 

1. Click the AI icon in the navigation bar 
2. Type your prompt in the text box
3. Attach any supporting files if needed 
4. Click Send to run the prompt 

Remember that Benchling AI performs best when your prompt is clear and specific. 

Attaching files helps Benchling AI match the format and level of detail that you need. Consider attaching: 

• A report template for the desired output structure
• A recent example to establish voice and formatting
• A file with specific constraints such as audience, due date, or data cut 

For example, to prepare for a lab meeting, you might attach your template and a recent example and then prompt: 

Attached are 1) the general template and 2) a recent example of my weekly lab reports. 
Review my Notebook entries from the last two weeks. 
Summarize the relevant data and outcomes of those entries following this format. 
Format the sections so that I can easily copy and paste them into PowerPoint slides. 
It will be particularly helpful if you can include citation links to entries with images or charts that I could include. 

Use skills 

If you have uploaded custom skills to your profile, then all of your custom skills and all available Benching-authored skills are surfaced automatically and the relevant skill(s) are applied throughout the conversation. However, if you want to use a specific skill, you can state that in the chat. 

Generate images

Benchling AI can generate scientific diagrams, workflow visualizations, and conceptual illustrations directly in your chat using Gemini.

To generate an image, start a chat and include a request like:

• "Create an image of…"
• "Generate a diagram of…"

Tips for better image generation

• Be specific about layout — for example, "left-to-right flow," "U-shaped diagram," or "branching paths that converge"
• Define visual style — for example, "use a yellow glow around the mouse to indicate irradiation"
• Include labeling requirements — for example, "label each arrow with the process name"
• Specify color and contrast — for example, "white background with dark text" for publication readability

Note: If your organization has style guidelines for generated images, add them to the tenant-level Assistant guidelines in the Tenant Admin Console.

Save outputs 

At the bottom of chat outputs, click Save as to save as: 

• A .docx Word file
• A new Notebook entry or template
• An addition appended to an existing Notebook entry

Review and export citations and plots

Benchling AI’s core tool set is similar to and interacts with the same tools you have access to in the Benchling UI. The Tool Details modal will show you the input and output of each step in Benchling AI’s process. If you would like to review the same data in the Benchling UI yourself you can do so by following these steps for the corresponding type of tool call:

Search Benchling, search Notebook entries, read Notebook, and look things up

These are all either search queries and their results or reading the contents of a Benchling Notebook Entry, its attachments, a Benchling Entity (sequence, custom entity) or other Benchling object (task).

• Search Queries will be formatted like Query: "CL001,CL002" 
  1. Replicate this by pasting each item in the comma separated list into Benchling search 
• Reading Notebook or Looking things up will typically return the Entry Title, EXP ID or both near the top of the input. If you see an API ID like etr_... bfi_... or seq_... you can: 
  1. Use Global search 
  2. Add Filters 
  3. Select API ID from the filters list and then search for one or more (comma separated) API IDs

Read attachment

These refer to files attached to your chat. Files attached to Benchling Notebook entries will be referenced in "Reading Notebook" citations. You can attach any of the following: 

• A report template the agent should follow
• A recent example (for voice, level of detail, and formatting) 
• Any constraints (audience, due date, slide length, data cut) 

Query the warehouse

These should be formatted like Query: "SELECT ..."

1. Create a new block in dashboard in the Insights application and replace the default query body with the query between the double quotation marks and then click "Run query" 
2. You can use SQL Writer to modify this query for other, similar use cases

"Create chart" for custom plots, charts, or graphs

The Input will contain the python code for a function named "llm_make_figure." This will include the data that will be incorporated in the graphic(s) itself, not necessarily the raw data (e.g. if the average values are being graphed, the code likely only contains the averages, not the datapoints from which they were derived). 

You can insert this function into any python context and perform actions on the figure including exporting to several image formats. 

For detailed instructions, see the next section.

Export custom plots from chat citations 

In a chat where you have created a custom chart or graph, you can export the interactive chart as a static image following these instructions:

1. To make adjustments to how the chart looks, ask Benchling AI a follow-up question in the chat thread asking it to make the changes you would like 
2. Once you have a visual you are satisfied with, click the Creating chart activity item in the Thinking section at the top of the chat response
3. The Input box will contain the python code for a function named llm_make_figure
   • This will include the data that will be incorporated in the graphic(s) itself, not necessarily the raw data e.g. if the average values are being graphed, the code likely only contains the averages, not the datapoints from which they were derived. 
4.  Insert this function into any python context and perform actions on the figure including exporting to several image formats

Export to .svg by Windows

1. Open a blank file in Notepad or another text editing or IDE application 
2. Copy and paste the "Input:" from the "Creating chart" activity
3. Replace return fig on the last row with fig.write_image("chart.svg")
   • Replace .svg with .png etc to save to another supported format
4. Open command prompt or powershell (or another application able to run python such as Visual Studio Code) 
5. Type python then "return" or "enter" (you should now see >>>)
6. Copy and paste in your modified code from above then "return" or "enter" until you again see see >>>
7. Type llm_make_figure() then "return" or "enter"
8. The image file should now be in the directory where you were when you started the terminal, by default typically your user's home folder (where Documents and Desktop are usually located)
9. To exit, type exit() or quit the command prompt app
    

Export to .svg by Mac, Linux

1. Open a blank file in Textedit or another text editing or IDE application.
2. Copy and paste the "Input:" from the "Creating chart" activity
3. Replace return fig on the last row with fig.write_image("chart.svg")
   • Replace .svg with .png etc to save to another supported format
4. Open Terminal (or another application able to run python such as Visual Studio Code) 
5. Type python then "return" or "enter" (you should now see >>>)
6. Copy and paste in your modified code from above then return or enter until you again see see >>>
7. Type llm_make_figure() then return or enter
8. The image file should now be in the directory where you were when you started the terminal, by default typically your user's home folder (where Documents and Desktop are usually located)
9. To exit, type exit() or quit the command prompt app

Export to .svg by Jupyter notebook 

This includes cloud-based options such as Google Colab

1. Create a new notebook
2. Create a new code block
3. You may need to run a few commands such as those below and run them first in a separate code block -- the error messages should guide you through this one-time setup
   • !pip install -U kaleido
   • import kaleido
   • kaleido.get_chrome_sync() 
4. Copy and paste the "Input:" from the "Creating chart" activity
5. Replace return fig on the last row with fig.write_image("chart.svg")
   • Replace .svg with .png etc to save to another supported format.
6. Click Run All or Run Cell
7. The File should now be available in your Files, by default in the /content/ directory 

Access previous chats 

Any query that you have made can be returned  to at a later time by following these steps: 

1. Click the AI icon in the navigation bar
2. Click on Chat history at the top of the chat window 
3. Scroll through the list of chats and click on the one you want to open to continue the conversation

Understand outputs 

Chat outputs include two types of references you can use to verify results: 

• Benchling object references – which appear as interactive chips that you click in order to navigate directly to the source or hover over to preview information

• Data citations – which appear as bracketed superscript links that you click to view the source context for that specific step in the agent’s process

Benchling AI with public data 

When using Benchling AI with public data search, you can ask questions across both your Benchling data and public sources. Benchling AI automatically crafts a query using your data, which is then used to query the internet. 

Note: Benchling AI with public data search is disabled by default. To enable, go to your AI settings in the tenant admin console. Once enabled, start a chat and toggle the option to include public data.

Connect external data with MCPs

Model Context Protocols (MCPs) allow Benchling AI to pull in data from external systems like Microsoft Teams, SharePoint, Slack, and other MCP-compatible platforms. Your admin must configure MCPs at the tenant level before you can connect them to your account.

Connect an MCP

1. Click the Connections icon (the last icon in the navigation bar)
2. Click the MCPs tab
3. Click the MCP you want to connect
4. Click Authenticate or Connect and complete authentication in the window that opens
5. Return to Benchling and finalize the MCP
6. Select which tools you want to enable from the options the admin has made available
7. Click Save

Benchling AI automatically determines which connected MCPs to query based on your prompt. You can't enable or disable specific MCPs on a per-chat basis, but you can return to the MCPs tab to adjust your settings at any time.

Manage connected MCPs

From the MCPs tab, you can:

• Configuration:  Adjust your personal settings, including which tools are enabled
• Toggle:  Turn the MCP on or off
• Disconnect:  Remove the MCP from your account entirely

Common use cases 

• Creating structured reports from a template [Start by uploading a template file with the desired output structure, examples linked at the bottom of this page]
  • Certificate of Analysis (CoA) or Certificate of Testing (CoT)
  • Technology Transfer Report
  • Non-clinical Study Report, including first drafts of IND enabling or BLA enabling reports like CTD CMC reports 
    • Example: "Using the attached file ‘Nonclinical Study Report.docx’ as a template, generate a study report for experiment EXP2500001. Fill each section of the template with data from Benchling. Make sure to provide citations for each piece of data."
  • General Research Study Report
    • Example: "Use the attached template to generate a report about the study in the "EBV Study" folder."  
•  Finding entities with specific relationships or missing connections
  • Example: "Identify all plasmids registered in the last 2 months that do not have linked stable cell lines. Create a table showing: Plasmid ID, Registration Date, Gene Insert, Selection Marker, and Intended Cell Line Type."
• Condensing complex experimental data into digestible insights
  • Example: "Summarize all experiments from Study ST042 conducted in the last 30 days. Include completion status, primary endpoints met (>90% viability threshold), any deviations from protocol, and remaining milestones. Format as an executive summary showing completed vs. planned experiments."
• Generating formatted documents from experimental data
  • Example: "Generate a non-clinical study report for experiment EXP2500001 including: 1) Study objective and rationale, 2) Test article information with lot numbers, 3) Methods section referencing SOPs, 4) Results with all raw data from linked Results tables, 5) Existing statistical analyses with p-values, 6) Conclusions addressing the primary hypothesis. Include all quality control checkpoints and flag any values outside acceptance criteria (CV > 15%)."
• Investigating experimental failures and identifying patterns
  • Example: "Analyze all failed transfection experiments (efficiency <20%) in Project X. Compare against successful transfections (efficiency >70%) from the same period, examining: cell passage number, DNA concentration, reagent lot numbers, operator, incubation time, and confluence percentage. Create a ranked list of variables most correlated with failure. What caused EXP2500001 to go wrong?"
• Evaluating data organization and access permissions
  • Example: "Audit all notebook entries and entities created in Q4 2024 for compliance issues. Identify: 1) Entries without project associations, 2) Entities missing required metadata fields (lot number, expiration date, creator), 3) Users with edit permissions on archived projects, 4) Sensitive data (patient/clinical) accessible to non-clinical team members. Output a prioritized action list with specific remediation steps."
• Exploring research directions and identifying knowledge gaps
  • Example: "Analyze all data from Project Y to identify research gaps. Map which combinations of our lead compounds and cell lines have been tested, what assays have been run (binding, toxicity, stability), and what concentrations have been evaluated. Create a matrix showing tested vs. untested conditions and recommend the top 5 experiments to run next based on existing trends and missing critical data."
• Generating workflow schematics
  • Example: “Create a diagram of our antibody purification workflow"
• Diagram a study or experiment:
  • Example: “Summarize the study described in EXP2500123. Include a diagram outlining the study design.”
• ERD
  • Example: “Make an image that shows the entity schemas in our registry and their relationships as an ERD.”
  • Example: “Summarize the lineage and links to key entries for purified antibody lot Benchlingimab 1.4-001 Produce an image with a diagram of these relationships. 
• Transcribe a protocol into a template
  • Example: "Create a notebook template based on the attached protocol. Include any appropriate result tables."
• Create a template from existing entries
  • Example: "Templatize the experiments conducted in EXPXXX and EXPXXX." 
• Transcribe an externally documented experiment
  • Example: "Transcribe the attached lab notebook file into a Benchling entry. Import any data into structured result tables."
• Append protocol steps from an external file to an existing entry
  • Example: "Transcribe the attached lab notebook file into a Benchling entry."
• Extract methods from a scientific publication
  • Example: "Based on the attached publication, draft an experiment protocol entry that captures the methods described. Include reagents, equipment, and step-by-step procedures. Format it as an actionable lab protocol that our team can follow to reproduce the experiment." 

Risks from AI mistakes

As with any AI tool, Benchling AI may make mistakes, so it is critical to verify the results to ensure the data is properly entered into the system. 

• All AI-based systems may make mistakes. Treat outputs as a draft and verify before sharing or saving to a Notebook
• The system may produce false positives, which may be misleading. Benchling AI will include citation links and chip links where possible to facilitate double-checking its responses
• Think of Benchling AI as a capable new teammate who just started working at your company. Use its speed to reduce searching and formatting work so you can focus on analysis and polish

Data protection & security

Benchling AI operates on behalf of the user who made the query. That means it has access to the same data as the user who made the query, not all of the data in Benchling. 

For more information about privacy and security for AI-powered features see the linked Data protection and security for AI at Benchling article. 

Frequently Asked Questions

• Can I choose which MCPs Benchling AI uses for specific questions?
  • Not today. All MCPs you've enabled in the MCP Connections page are available to Benchling AI, which automatically determines which systems to query based on your prompt. You cannot enable or disable specific MCPs on a per-chat basis through the chat UI. Though you can return to the MCP tab to enable or disable MCPs at any time. 
• ​​Why am I getting an authentication error?
  • If you encounter authentication errors when setting up your MCP, the external system may require pre-registration of the MCP Client (OAuth client) on the MCP server side. Contact Benchling Support for assistance with authentication issues.