Elnora plans experiments, writes protocols, searches the literature and analyses results, helping your team discover faster inside the tools you already use.
Cortical neuron protocol
I need cortical neurons from iPSCs in three weeks. Write me the protocol.
AGENT ACTIVITY
54sAnalysing your request
Loading fair-data-protocols
Saving protocol(0.2s)
I have drafted and saved the 21-day NGN2 differentiation protocol, written to your lab's own conventions:
Seeding density: 1.5 × 10⁵ cells/cm², capped at 1.8 × 10⁵.
Ara-C timing: Scheduled for the Day 4 media change.
Feeding schedule: Half-media changes every 2 to 3 days.
Laminin: Included at 1 µg/mL to prevent detachment.
QC checkpoint: Day 14, staining for MAP2 and vGLUT1, >90% target.
Tracking: Always record the Matrigel lot number.
The protocol is saved in your Knowledge Base as ngn2-cortical-neuron-differentiation.md
Failed runs review
Here are runs 014 and 016. Both failed. Figure out why and stop it from happening again.
AGENT ACTIVITY
44sReading run 014 and run 016(0.8s)
Comparing seeding density, Ara-C timing, and feed schedule(32s)
Writing standing decisions to memory(0.4s)
Both runs seeded above the safe range and missed the Day 4 Ara-C window. I have written these as standing rules for every future run:
Seeding: 1.5 × 10⁵ cells/cm², never above 1.8 × 10⁵.
Ara-C: Day 4, never later.
Feeding: Half-media every 2 to 3 days.
Laminin: 1 µg/mL.
Checkpoint: Day 14, MAP2 at 90% or above.
Tracking: Record the Matrigel lot on every run.
Patent landscape, NGN2
Is anyone patenting NGN2 based neuron differentiation methods? Show me the key filings and who owns them.
AGENT ACTIVITY
9m 40sLoading patent-search(0.04s)
Patent search for 'NGN2 neuron differentiation'(9.8s)
Searching USPTO(0.15s)
Searching EPO for 'ta=Neurogenin and ta=neuron'(0.15s)
Searching Google Patents for 'Neurogenin neuron'(5.7s)
Across USPTO, EPO and Google Patents the landscape rests on one foundational family covering transcription factor conversion, with the NGN2 protocol for differentiating iPSCs into neurons in wide open use.
US 9,057,053 B2 and US 9,822,338 B2: Direct conversion of cells to cells of other lineages.
Assignee: The Board of Trustees of the Leland Stanford Junior University.
Scope: the granted claims name Ascl1, Brn2, Myt1l and NeuroD, which is the basis the method builds on.
Literature scan, NGN2
Pull everything recent on NGN2 induction efficiency and tell me what is worth reading.
AGENT ACTIVITY
2m 14sAnalysing your request
Loading external-search
Searching PubMed for recent NGN2 induction protocols(0.2s)
Searching OpenAlex, 20 sources(0.8s)
Searching the web, 8 papers(0.9s)
The most widely adopted method for generating mature cortical neurons in 21 days relies on doxycycline-inducible overexpression of Neurogenin-2, originally Zhang et al., 2013, and optimised by Shan et al., 2024 and Fernandopulle et al., 2019.
Resources used (2)
Why did the transfection fail
My transfection into hippocampal neurons keeps failing. What should I check first?
AGENT ACTIVITY
54sSearching file contents(0.08s)
Reading file hippocampal-neuron-gabaa-transfection.md(0.17s)
Hippocampal neurons are stubborn to transfect, so separate toxicity, where the cells die, from low efficiency, where they survive with little expression. Your own transfection protocol in the knowledge base points at the likely causes.
Conditioned media: Lipofectamine 2000 is toxic to primary neurons. Your own protocol caps exposure at 1 to 2 hours and returns the saved conditioned media after the wash.
Neuron age: efficiency falls off sharply as the culture matures.
DNA to reagent ratio: the protocol you already run specifies it.
Answered from your own file, hippocampal-neuron-gabaa-transfection.md
Elnora plans the steps, does the work, and comes back when it needs a decision from you.
Give it your protocols, results and notes. What comes back uses your reagents, your instruments and the rules your team follows.
Open it in a browser and point it at the files your team already keeps, so you can start with what you have today.
Your own agents call Elnora as a tool over a hosted MCP server, a REST API or a command line.
BUILT FOR
Elnora works across life sciences, from biology and chemistry to materials, electrochemistry and physics. As long as you are performing experiments in the lab, Elnora is for you.
RESEARCH
Every claim comes back with the source behind it, from PubMed, bioRxiv, UniProt, ChEMBL, patents and 171 scientific databases in all.
Learn moreAGENT ACTIVITY
2m 14sAnalysing your request
Loading external-search
Searching PubMed for recent NGN2 induction protocols(0.2s)
Searching OpenAlex, 20 sources(0.8s)
Searching Europe PMC(0.4s)
Searching the web, 8 papers(0.9s)
Resources used (2)
MEMORY
Everything it writes starts from your own protocols, results and notes, kept in one place, private to you or shared with the team.
Learn morengn2-cortical-neuron-differentiation.md
QC Checkpoint 4 (Day 14): Immunofluorescence should confirm >90% positivity for MAP2, NeuN, and vGLUT1.
ipsc-cortical-neuron-ngn2-21day.md
QC Checkpoint 5 (Day 21): Patch-clamp or MEA recordings should detect spontaneous action potentials.
NEGATIVE DATA
The next experiment it plans starts from what went wrong last time and names the variable to change, because you gave it every run, the failures included.
Learn moreRun 014
Failedseeded 2.4 × 10⁵ cells/cm²
sheet detachment day 4, MAP2 41% at day 14
Run 016
Failedseeded 1.8 × 10⁵, Ara-C day 4
network thinning day 10, MAP2 68% at day 14
FILED IN THE KNOWLEDGE BASE
Seeding density: 1.5 × 10⁵ cells/cm², never above 1.8 × 10⁵
Ara-C: day 4 media change, never later
Feeding: half-media every 2 to 3 days
Laminin: 1 µg/mL
QC checkpoint: day 14, MAP2 at 90% or above
Tracking: record the Matrigel lot number
PROTOCOLS
Elnora writes the steps, reagents, volumes and QC checkpoints, prints each catalogue number once it resolves to a live product page, and revises from your feedback until it is ready to run.
Learn morev1 of 1
VERSION CONTROL
Every change to a file is kept with who made it, so you can always read the version that came before.
Learn moreANALYSIS
Hand it a plate of readings and it picks the right test, checks the assumptions, draws the figure and saves it where the next person finds it.
Learn moreGABA-A α5 inhibitor, plate 7
Four parameter logistic fit, R squared 0.992. A Dunnett test puts the lowest observed effect level at 0.3 µM.
THE WORK
It takes the work from the question to the file your team can use, and asks when it needs you.
Conditions, controls, replicates and the sample size the comparison actually needs.
Curves from plate readings, potency with confidence intervals, compared against on-plate controls.
Plate readers and potentiostats, read straight from the instrument's own export.
A document, a deck or a workbook you can edit, with the methods and figures in it.
SECURITY
ISO/IEC 27001:2022 certified and SOC 2 Type 2 attested. Your data is never used to train models, every organisation runs isolated, and every protocol and result belongs to you.
Learn moreAn online panel on AI literacy in healthcare, with Carmen Kivisild among speakers from clinical care, public health, data and informatics, on what professionals need to know about AI and which skills to build.
Carmen Kivisild spoke at the LRIG Bay Area fall seminar on connecting AI agents to robotic and cloud labs, with the time and cost measured against the traditional path.
The Cypher team spent the day with Elnora at Altitude Lab in Salt Lake City, talking about AI, shared value for customers and projects the two companies could build together.
